// CHECK-LABEL: @test_svabd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svabd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svabd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svabd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svabd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svabd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svabd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svabd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svabd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svabd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svabd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svabd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svabd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
return SVE_ACLE_FUNC(svabd,_f16,_x,)(pg, op1, op2);
}
+//
// CHECK-LABEL: @test_svabd_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-LABEL: @test_svabd_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svabd_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svabd_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svabd_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svabd_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svacge_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z17test_svacge_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svacge_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z17test_svacge_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svacgt_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z17test_svacgt_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svacgt_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z17test_svacgt_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svacle_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z17test_svacle_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svacle_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z17test_svacle_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svaclt_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z17test_svaclt_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svaclt_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z17test_svaclt_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svadd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svadd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svadd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svadd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svadd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svadd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svadd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svadd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svadd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svadd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svadd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svadd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svadd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svadd_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svadd_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svadd_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svadd_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svand_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svand_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svand_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svand_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svand_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svand_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svand_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svand_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svand_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svand_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svand_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svand_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svand_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svand_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svand_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svand_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svand_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svand_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svand_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svasr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s64_zu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svasr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s64_mu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svasr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s64_xu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svasr_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svasr_n_s8_zu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svasr_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s16_zu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svasr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s32_zu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svasr_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svasr_n_s8_mu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svasr_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s16_mu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svasr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s32_mu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svasr_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svasr_n_s8_xu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svasr_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s16_xu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svasr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s32_xu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svasr_wide_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svasr_wide_n_s8_mu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svasr_wide_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s16_mu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svasr_wide_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s32_mu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svasr_wide_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z22test_svasr_wide_n_s8_zu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svasr_wide_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s16_zu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svasr_wide_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s32_zu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svasr_wide_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svasr_wide_n_s8_xu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svasr_wide_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s16_xu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svasr_wide_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s32_xu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_bfdot_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_bfdot_n_f32u13__SVFloat32_tu14__SVBFloat16_tu6__bf16(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_bfmlalb_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_bfmlalb_n_f32u13__SVFloat32_tu14__SVBFloat16_tu6__bf16(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_bfmlalt_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_bfmlalt_n_f32u13__SVFloat32_tu14__SVBFloat16_tu6__bf16(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CHECK-LABEL: @test_svbic_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svbic_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svbic_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svbic_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svbic_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svbic_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svbic_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svbic_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svbic_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbic_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbic_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbic_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbic_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svbic_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svcmpeq_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpeq_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpeq_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpeq_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpeq_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpeq_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpeq_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpeq_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpeq_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpeq_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpeq_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpeq.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpeq.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpeq_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpeq.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpeq.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpeq_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpeq.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpeq.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpeq_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpeq_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpeq_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpeq_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpeq_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpeq_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpge_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpge_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpge_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpge_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpge_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpge_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_wide_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpge_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpge_wide_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpge_wide_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpgt_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpgt_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpgt_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpgt_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpgt_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpgt_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_wide_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpgt_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpgt_wide_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpgt_wide_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmple_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmple_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmple_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmple_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmple.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmple_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmple.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmple_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmple.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmple.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmple.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmple.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_wide_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpls.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmple_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpls.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmple_wide_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpls.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpls.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmple_wide_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpls.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpls.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmplt_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmplt_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmplt_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmplt_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmplt_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmplt_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_wide_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplo.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmplt_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplo.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmplt_wide_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplo.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplo.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmplt_wide_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplo.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplo.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpne_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpne_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpne_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpne_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpne_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svcmpne_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpne_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpne_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svcmpne_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpuo_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpuo.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpuo_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpuo.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpuo_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpuo.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpuo_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpuo.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svcmpuo_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpuo.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svcmpuo_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpuo.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
// CHECK-LABEL: @test_svdiv_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svdiv_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svdiv_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svdiv_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svdiv_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svdiv_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svdiv_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svdiv_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svdiv_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svdivr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svdivr_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svdivr_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svdivr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svdivr_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svdivr_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svdivr_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svdivr_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svdot_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdot_n_s32u11__SVInt32_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svdot_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdot_n_s64u11__SVInt64_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svdot_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdot_n_u32u12__SVUint32_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svdot_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdot_n_u64u12__SVUint64_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_n_bf16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 8 x bfloat> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z17test_svdup_n_bf16u6__bf16(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 8 x bfloat> [[TMP0]]
//
svbfloat16_t test_svdup_n_bf16(bfloat16_t op) {
// CHECK-LABEL: @test_svdup_lane_bf16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.tbl.nxv8bf16(<vscale x 8 x bfloat> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x bfloat> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svdup_lane_bf16u14__SVBFloat16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.tbl.nxv8bf16(<vscale x 8 x bfloat> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x bfloat> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svdup_n_s8a(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
//
svint8_t test_svdup_n_s8(int8_t op)
// CHECK-LABEL: @test_svdup_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_s16s(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
//
svint16_t test_svdup_n_s16(int16_t op)
// CHECK-LABEL: @test_svdup_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_s32i(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
svint32_t test_svdup_n_s32(int32_t op)
// CHECK-LABEL: @test_svdup_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_s64l(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
//
svint64_t test_svdup_n_s64(int64_t op)
// CHECK-LABEL: @test_svdup_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svdup_n_u8h(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
//
svuint8_t test_svdup_n_u8(uint8_t op)
// CHECK-LABEL: @test_svdup_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_u16t(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
//
svuint16_t test_svdup_n_u16(uint16_t op)
// CHECK-LABEL: @test_svdup_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_u32j(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
svuint32_t test_svdup_n_u32(uint32_t op)
// CHECK-LABEL: @test_svdup_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_u64m(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
//
svuint64_t test_svdup_n_u64(uint64_t op)
// CHECK-LABEL: @test_svdup_n_f16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_f16Dh(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP0]]
//
svfloat16_t test_svdup_n_f16(float16_t op)
// CHECK-LABEL: @test_svdup_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_f32f(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP0]]
//
svfloat32_t test_svdup_n_f32(float32_t op)
// CHECK-LABEL: @test_svdup_n_f64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_f64d(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP0]]
//
svfloat64_t test_svdup_n_f64(float64_t op)
// CHECK-LABEL: @test_svdup_lane_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svdup_lane_s8u10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_lane_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_s16u11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_lane_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_s32u11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_lane_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_s64u11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_lane_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svdup_lane_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_lane_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_lane_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_lane_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_lane_f16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.tbl.nxv8f16(<vscale x 8 x half> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_f16u13__SVFloat16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.tbl.nxv8f16(<vscale x 8 x half> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_lane_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.tbl.nxv4f32(<vscale x 4 x float> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_f32u13__SVFloat32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.tbl.nxv4f32(<vscale x 4 x float> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_lane_f64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.tbl.nxv2f64(<vscale x 2 x double> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_f64u13__SVFloat64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.tbl.nxv2f64(<vscale x 2 x double> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_n_b8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.dup.x.nxv16i1(i1 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i1> [[DOTSPLATINSERT]], <vscale x 16 x i1> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svdup_n_b8b(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.dup.x.nxv16i1(i1 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i1> [[DOTSPLATINSERT]], <vscale x 16 x i1> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP0]]
//
svbool_t test_svdup_n_b8(bool op)
// CHECK-LABEL: @test_svdup_n_b16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.dup.x.nxv8i1(i1 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i1> [[DOTSPLATINSERT]], <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_b16b(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.dup.x.nxv8i1(i1 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i1> [[DOTSPLATINSERT]], <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_n_b32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.dup.x.nxv4i1(i1 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i1> [[DOTSPLATINSERT]], <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_b32b(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.dup.x.nxv4i1(i1 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i1> [[DOTSPLATINSERT]], <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-LABEL: @test_svdup_n_b64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.dup.x.nxv2i1(i1 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i1> [[DOTSPLATINSERT]], <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_b64b(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.dup.x.nxv2i1(i1 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i1> [[DOTSPLATINSERT]], <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CHECK-NEXT: [[TMP16:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
// CHECK-NEXT: [[TMP17:%.*]] = call <vscale x 16 x i8> @llvm.experimental.vector.insert.nxv16i8.v16i8(<vscale x 16 x i8> undef, <16 x i8> [[TMP15]], i64 0)
// CHECK-NEXT: [[TMP18:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dupq.lane.nxv16i8(<vscale x 16 x i8> [[TMP17]], i64 0)
-// CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
-// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP20]]
+// CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP19]]
//
// CPP-CHECK-LABEL: @_Z16test_svdupq_n_b8bbbbbbbbbbbbbbbb(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP16:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
// CPP-CHECK-NEXT: [[TMP17:%.*]] = call <vscale x 16 x i8> @llvm.experimental.vector.insert.nxv16i8.v16i8(<vscale x 16 x i8> undef, <16 x i8> [[TMP15]], i64 0)
// CPP-CHECK-NEXT: [[TMP18:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dupq.lane.nxv16i8(<vscale x 16 x i8> [[TMP17]], i64 0)
-// CPP-CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
-// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP20]]
+// CPP-CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP19]]
//
svbool_t test_svdupq_n_b8(bool x0, bool x1, bool x2, bool x3,
bool x4, bool x5, bool x6, bool x7,
// CHECK-NEXT: [[TMP16:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
// CHECK-NEXT: [[TMP17:%.*]] = call <vscale x 8 x i16> @llvm.experimental.vector.insert.nxv8i16.v8i16(<vscale x 8 x i16> undef, <8 x i16> [[TMP15]], i64 0)
// CHECK-NEXT: [[TMP18:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dupq.lane.nxv8i16(<vscale x 8 x i16> [[TMP17]], i64 0)
-// CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
-// CHECK-NEXT: [[TMP21:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP20]])
-// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP21]]
+// CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP19]])
+// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP20]]
//
// CPP-CHECK-LABEL: @_Z17test_svdupq_n_b16bbbbbbbb(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP16:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
// CPP-CHECK-NEXT: [[TMP17:%.*]] = call <vscale x 8 x i16> @llvm.experimental.vector.insert.nxv8i16.v8i16(<vscale x 8 x i16> undef, <8 x i16> [[TMP15]], i64 0)
// CPP-CHECK-NEXT: [[TMP18:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dupq.lane.nxv8i16(<vscale x 8 x i16> [[TMP17]], i64 0)
-// CPP-CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
-// CPP-CHECK-NEXT: [[TMP21:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP20]])
-// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP21]]
+// CPP-CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP19]])
+// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP20]]
//
svbool_t test_svdupq_n_b16(bool x0, bool x1, bool x2, bool x3,
bool x4, bool x5, bool x6, bool x7)
// CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
// CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v4i32(<vscale x 4 x i32> undef, <4 x i32> [[TMP7]], i64 0)
// CHECK-NEXT: [[TMP10:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dupq.lane.nxv4i32(<vscale x 4 x i32> [[TMP9]], i64 0)
-// CHECK-NEXT: [[TMP11:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> [[TMP11]])
-// CHECK-NEXT: [[TMP13:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP12]])
-// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP13]]
+// CHECK-NEXT: [[TMP11:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP11]])
+// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP12]]
//
// CPP-CHECK-LABEL: @_Z17test_svdupq_n_b32bbbb(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
// CPP-CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v4i32(<vscale x 4 x i32> undef, <4 x i32> [[TMP7]], i64 0)
// CPP-CHECK-NEXT: [[TMP10:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dupq.lane.nxv4i32(<vscale x 4 x i32> [[TMP9]], i64 0)
-// CPP-CHECK-NEXT: [[TMP11:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> [[TMP11]])
-// CPP-CHECK-NEXT: [[TMP13:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP12]])
-// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP13]]
+// CPP-CHECK-NEXT: [[TMP11:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP11]])
+// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP12]]
//
svbool_t test_svdupq_n_b32(bool x0, bool x1, bool x2, bool x3)
{
// CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
// CHECK-NEXT: [[TMP5:%.*]] = call <vscale x 2 x i64> @llvm.experimental.vector.insert.nxv2i64.v2i64(<vscale x 2 x i64> undef, <2 x i64> [[TMP3]], i64 0)
// CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dupq.lane.nxv2i64(<vscale x 2 x i64> [[TMP5]], i64 0)
-// CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]])
-// CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP8]])
-// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP9]]
+// CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP7]])
+// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP8]]
//
// CPP-CHECK-LABEL: @_Z17test_svdupq_n_b64bb(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
// CPP-CHECK-NEXT: [[TMP5:%.*]] = call <vscale x 2 x i64> @llvm.experimental.vector.insert.nxv2i64.v2i64(<vscale x 2 x i64> undef, <2 x i64> [[TMP3]], i64 0)
// CPP-CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dupq.lane.nxv2i64(<vscale x 2 x i64> [[TMP5]], i64 0)
-// CPP-CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]])
-// CPP-CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP8]])
-// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP9]]
+// CPP-CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP7]])
+// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP8]]
//
svbool_t test_svdupq_n_b64(bool x0, bool x1)
{
// CHECK-LABEL: @test_sveor_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_sveor_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_sveor_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_sveor_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_sveor_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_sveor_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_sveor_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_sveor_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_sveor_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_sveor_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_sveor_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_sveor_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_sveor_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_sveor_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svlsl_wide_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsl_wide_n_s8_mu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svlsl_wide_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s16_mu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svlsl_wide_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s32_mu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svlsl_wide_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsl_wide_n_s8_zu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svlsl_wide_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s16_zu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svlsl_wide_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s32_zu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svlsl_wide_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsl_wide_n_s8_xu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svlsl_wide_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s16_xu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svlsl_wide_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s32_xu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svlsr_wide_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsr_wide_n_u8_mu10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svlsr_wide_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u16_mu10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svlsr_wide_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u32_mu10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svlsr_wide_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsr_wide_n_u8_zu10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svlsr_wide_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u16_zu10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svlsr_wide_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u32_zu10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svlsr_wide_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsr_wide_n_u8_xu10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svlsr_wide_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u16_xu10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svlsr_wide_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u32_xu10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmad_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmad_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmad_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmad_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmad_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmad_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmad_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmad_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmad_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmad_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmad_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmad_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svmad_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svmad_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svmad_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmad_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmax_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmax_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmax_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmax_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmax_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmax_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmax_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmax_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmax_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmax_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmax_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmax_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svmax_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svmax_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svmax_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmax_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmaxnm_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svmaxnm_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svmaxnm_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svmaxnm_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmaxnm_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmaxnm_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmaxnm_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmaxnm_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmaxnm_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmin_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmin_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmin_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmin_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmin_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmin_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmin_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmin_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmin_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmin_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmin_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmin_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svmin_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svmin_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svmin_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmin_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svminnm_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svminnm_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svminnm_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svminnm_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svminnm_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svminnm_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svminnm_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svminnm_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svminnm_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmla_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmla_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmla_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmla_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmla_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmla_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmla_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmla_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmla_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmla_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmla_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmla_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svmla_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svmla_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svmla_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmla_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmls_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmls_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmls_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmls_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmls_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmls_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmls_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmls_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmls_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmls_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmls_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmls_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svmls_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svmls_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svmls_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmls_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmsb_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmsb_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmsb_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmsb_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmsb_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmsb_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmsb_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmsb_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmsb_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmsb_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmsb_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmsb_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svmsb_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svmsb_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svmsb_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmsb_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmul_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmul_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmul_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmul_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmul_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmul_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmul_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmul_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmul_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmul_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmul_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmul_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svmul_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svmul_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svmul_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmul_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmulh_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmulh_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmulh_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmulh_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svmulh_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svmulh_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svmulh_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svmulh_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmulh_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmulh_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmulh_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svmulh_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svmulh_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svmulx_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svmulx_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svmulx_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svmulx_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmulx_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmulx_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svmulx_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svmulx_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svmulx_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svnmad_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svnmad_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svnmad_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svnmad_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svnmad_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svnmad_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svnmad_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svnmad_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svnmad_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svnmla_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svnmla_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svnmla_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svnmla_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svnmla_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svnmla_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svnmla_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svnmla_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svnmla_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svnmls_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svnmls_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svnmls_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svnmls_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svnmls_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svnmls_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svnmls_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svnmls_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svnmls_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svnmsb_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svnmsb_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svnmsb_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svnmsb_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svnmsb_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svnmsb_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svnmsb_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svnmsb_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svnmsb_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svorr_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svorr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svorr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svorr_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svorr_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svorr_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svorr_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svorr_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svorr_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svorr_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svorr_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svorr_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svorr_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svqadd_n_s8u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svqadd_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svqsub_n_s8u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svqsub_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svscale_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f16_zu10__SVBool_tu13__SVFloat16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svscale_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f32_zu10__SVBool_tu13__SVFloat32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svscale_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f64_zu10__SVBool_tu13__SVFloat64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svscale_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f16_mu10__SVBool_tu13__SVFloat16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svscale_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f32_mu10__SVBool_tu13__SVFloat32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svscale_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f64_mu10__SVBool_tu13__SVFloat64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svscale_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f16_xu10__SVBool_tu13__SVFloat16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svscale_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f32_xu10__SVBool_tu13__SVFloat32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svscale_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f64_xu10__SVBool_tu13__SVFloat64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svsub_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svsub_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svsub_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svsub_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svsub_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svsub_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svsub_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svsub_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsub_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsub_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsub_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsub_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svsub_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svsub_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svsub_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svsub_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svsubr_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svsubr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svsubr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svsubr_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svsubr_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svsubr_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svsubr_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svsubr_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsubr_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsubr_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsubr_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsubr_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
// CHECK-LABEL: @test_svsubr_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
// CHECK-LABEL: @test_svsubr_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
// CHECK-LABEL: @test_svsubr_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CHECK-LABEL: @test_svsubr_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -target-feature +i8mm -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -target-feature +i8mm -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -target-feature +i8mm -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsudot_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[Y:%.*]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsudot_n_s32u11__SVInt32_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[Y:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -target-feature +i8mm -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -target-feature +i8mm -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -target-feature +i8mm -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svusdot_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[Y:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svusdot_n_s32u11__SVInt32_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[Y:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svaba_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.saba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svaba_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.saba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svaba_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaba_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaba_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svaba_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uaba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svaba_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uaba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svaba_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaba_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaba_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svabalb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svabalb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svabalb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svabalb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svabalb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svabalb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svabalt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svabalt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svabalt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svabalt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svabalt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svabalt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svabdlb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svabdlb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svabdlb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svabdlb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svabdlb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svabdlb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svabdlt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svabdlt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svabdlt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svabdlt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svabdlt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svabdlt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svadclb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svadclb_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svadclb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svadclb_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svadclt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svadclt_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svadclt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svadclt_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svaddhnb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svaddhnb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddhnb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddhnb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svaddhnb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddhnb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svaddhnt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_s16u10__SVInt8_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svaddhnt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_s32u11__SVInt16_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddhnt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_s64u11__SVInt32_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddhnt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_u16u11__SVUint8_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svaddhnt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_u32u12__SVUint16_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddhnt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_u64u12__SVUint32_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svaddlb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svaddlbt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlbt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddlbt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlbt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlbt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlbt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddlbt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlbt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlbt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlbt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddlbt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlbt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svaddlt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddlt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svaddwb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_s16u11__SVInt16_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddwb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_s32u11__SVInt32_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddwb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_s64u11__SVInt64_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svaddwb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_u16u12__SVUint16_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddwb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_u32u12__SVUint32_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddwb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_u64u12__SVUint64_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svaddwt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_s16u11__SVInt16_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddwt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_s32u11__SVInt32_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddwt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_s64u11__SVInt64_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svaddwt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_u16u12__SVUint16_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svaddwt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_u32u12__SVUint32_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svaddwt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_u64u12__SVUint64_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svbcax_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbcax_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbcax_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbcax_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbcax_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svbcax_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbcax_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbcax_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbcax_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbcax_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svbdep_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bdep.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbdep_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bdep.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbdep_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bdep.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbdep_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bdep.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbdep_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bdep.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbdep_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bdep.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbdep_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bdep.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbdep_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bdep.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svbext_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bext.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbext_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bext.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbext_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bext.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbext_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bext.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbext_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bext.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbext_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bext.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbext_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bext.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbext_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bext.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svbgrp_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bgrp.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbgrp_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bgrp.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbgrp_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bgrp.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbgrp_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bgrp.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbgrp_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bgrp.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbgrp_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bgrp.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbgrp_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bgrp.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbgrp_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bgrp.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svbsl_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svbsl_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svbsl_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svbsl1n_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbsl1n_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl1n_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl1n_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl1n_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl1n_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbsl1n_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl1n_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl1n_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl1n_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svbsl2n_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbsl2n_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl2n_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl2n_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl2n_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl2n_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbsl2n_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl2n_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl2n_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svbsl2n_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_sveor3_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_sveor3_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_sveor3_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_sveor3_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_sveor3_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_sveor3_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_sveor3_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_sveor3_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_sveor3_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_sveor3_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_sveorbt_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveorbt_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_sveorbt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_sveorbt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_sveorbt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_sveorbt_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveorbt_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_sveorbt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_sveorbt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_sveorbt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_sveortb_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveortb_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_sveortb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_sveortb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_sveortb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_sveortb_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveortb_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_sveortb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_sveortb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_sveortb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svhadd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhadd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhadd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svhadd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svhadd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svhadd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svhadd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svhadd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svhadd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svhadd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svhadd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhadd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhadd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhadd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svhsub_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svhsub_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svhsub_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svhsub_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svhsub_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svhsub_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svhsub_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svhsub_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svhsub_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhsub_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhsub_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhsub_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhsub_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhsub_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svhsubr_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svhsubr_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svhsubr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svhsubr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svhsubr_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svhsubr_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svhsubr_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svhsubr_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svhsubr_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhsubr_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhsubr_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhsubr_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svhsubr_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svhsubr_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svmlalb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalb_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svmlalt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svmlalt_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svmlslb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslb_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svmlslt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svmlslt_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svmullb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmullb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmullb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svmullb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmullb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmullb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svmullt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmullt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmullt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svmullt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svmullt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svmullt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svnbsl_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svnbsl_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svnbsl_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svnbsl_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svnbsl_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svnbsl_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svnbsl_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svnbsl_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svnbsl_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svnbsl_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svpmul_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmul.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svpmul_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmul.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svpmullb_pair_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z23test_svpmullb_pair_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svpmullb_pair_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z24test_svpmullb_pair_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svpmullb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svpmullb_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
// CHECK-LABEL: @test_svpmullb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svpmullb_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svpmullb_pair_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullb.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z24test_svpmullb_pair_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullb.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svpmullt_pair_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z23test_svpmullt_pair_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svpmullt_pair_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z24test_svpmullt_pair_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svpmullt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svpmullt_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
// CHECK-LABEL: @test_svpmullt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svpmullt_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svpmullt_pair_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullt.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z24test_svpmullt_pair_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullt.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqadd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svqadd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svqadd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svqadd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svqadd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svqadd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svqadd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svqadd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svqadd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqadd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqadd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqdmlalb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalb_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlalb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalb_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlalb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalb_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqdmlalbt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlalbt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlalbt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlalbt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlalbt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlalbt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqdmlalt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlalt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlalt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqdmlslb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslb_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlslb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslb_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlslb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslb_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqdmlslbt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlslbt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlslbt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlslbt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlslbt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlslbt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqdmlslt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlslt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmlslt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqdmulh_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqdmulh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqdmulh_n_s8u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqdmulh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmulh_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmulh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqdmulh_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmulh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmulh_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmulh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqdmulh_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmulh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmulh_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmulh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqdmulh_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmulh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqdmullb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmullb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmullb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqdmullt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmullt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqdmullt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqrdmlah_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlah.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrdmlah_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlah.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqrdmlah_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlah.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlah_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlah.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqrdmlah_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlah.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlah_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlah.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqrdmlah_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlah.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlah_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlah.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqrdmlsh_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlsh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrdmlsh_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlsh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqrdmlsh_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlsh_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqrdmlsh_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlsh_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqrdmlsh_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlsh_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqrdmulh_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmulh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrdmulh_n_s8u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmulh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqrdmulh_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmulh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmulh_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmulh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svqrdmulh_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmulh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmulh_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmulh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svqrdmulh_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmulh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmulh_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmulh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqrshl_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svqrshl_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svqrshl_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svqrshl_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svqrshl_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_u8_zu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svqrshl_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u16_zu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svqrshl_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u32_zu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svqrshl_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u64_zu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svqrshl_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqrshl_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_u8_mu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqrshl_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u16_mu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u32_mu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u64_mu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqrshl_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_u8_xu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqrshl_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u16_xu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u32_xu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqrshl_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u64_xu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqshl_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svqshl_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svqshl_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svqshl_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svqshl_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_u8_zu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svqshl_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u16_zu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svqshl_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u32_zu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svqshl_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u64_zu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svqshl_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqshl_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_u8_mu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqshl_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u16_mu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u32_mu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u64_mu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqshl_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_u8_xu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqshl_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u16_xu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u32_xu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqshl_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u64_xu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqsub_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svqsub_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svqsub_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svqsub_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svqsub_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svqsub_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svqsub_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svqsub_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svqsub_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqsub_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqsub_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svqsubr_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svqsubr_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svqsubr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svqsubr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svqsubr_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svqsubr_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svqsubr_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svqsubr_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svqsubr_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqsubr_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqsubr_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqsubr_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svqsubr_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svqsubr_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svraddhnb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svraddhnb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svraddhnb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svraddhnb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svraddhnb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svraddhnb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svraddhnt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_s16u10__SVInt8_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svraddhnt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_s32u11__SVInt16_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svraddhnt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_s64u11__SVInt32_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svraddhnt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_u16u11__SVUint8_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svraddhnt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_u32u12__SVUint16_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svraddhnt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_u64u12__SVUint32_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svrhadd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrhadd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrhadd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svrhadd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svrhadd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svrhadd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svrhadd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svrhadd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svrhadd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svrhadd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svrhadd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrhadd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrhadd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svrhadd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svrshl_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svrshl_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svrshl_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svrshl_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svrshl_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_u8_zu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svrshl_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u16_zu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svrshl_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u32_zu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svrshl_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u64_zu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svrshl_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrshl_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_u8_mu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrshl_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u16_mu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u32_mu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u64_mu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrshl_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_u8_xu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrshl_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u16_xu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u32_xu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svrshl_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u64_xu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svrsubhnb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrsubhnb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svrsubhnb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svrsubhnb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrsubhnb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svrsubhnb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svrsubhnt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_s16u10__SVInt8_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrsubhnt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_s32u11__SVInt16_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svrsubhnt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_s64u11__SVInt32_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svrsubhnt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_u16u11__SVUint8_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svrsubhnt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_u32u12__SVUint16_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svrsubhnt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_u64u12__SVUint32_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsbclb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sbclb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsbclb_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sbclb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsbclb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sbclb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsbclb_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sbclb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsbclt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sbclt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsbclt_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sbclt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsbclt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sbclt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsbclt_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sbclt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsqadd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsqadd_n_u8_mu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsqadd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u16_mu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svsqadd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u32_mu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svsqadd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u64_mu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svsqadd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsqadd_n_u8_zu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svsqadd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u16_zu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svsqadd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u32_zu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svsqadd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u64_zu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svsqadd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsqadd_n_u8_xu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsqadd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u16_xu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svsqadd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u32_xu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svsqadd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u64_xu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsubhnb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsubhnb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsubhnb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsubhnb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsubhnb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsubhnb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsubhnt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_s16u10__SVInt8_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsubhnt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_s32u11__SVInt16_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsubhnt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_s64u11__SVInt32_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsubhnt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_u16u11__SVUint8_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svsubhnt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_u32u12__SVUint16_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsubhnt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_u64u12__SVUint32_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsublb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsublb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsublb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svsublb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usublb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usublb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsublb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usublb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usublb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsublb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usublb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usublb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsublbt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublbt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsublbt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublbt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsublbt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublbt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsublbt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublbt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsublbt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublbt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsublbt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublbt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsublt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsublt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsublt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svsublt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usublt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usublt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsublt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usublt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usublt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsublt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usublt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usublt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsubltb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubltb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubltb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubltb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsubltb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubltb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubltb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubltb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsubltb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubltb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubltb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubltb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsubwb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_s16u11__SVInt16_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsubwb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_s32u11__SVInt32_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsubwb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_s64u11__SVInt64_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svsubwb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usubwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_u16u12__SVUint16_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usubwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsubwb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usubwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_u32u12__SVUint32_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usubwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsubwb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usubwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_u64u12__SVUint64_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usubwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svsubwt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_s16u11__SVInt16_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsubwt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_s32u11__SVInt32_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsubwt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_s64u11__SVInt64_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CHECK-LABEL: @test_svsubwt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usubwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_u16u12__SVUint16_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usubwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CHECK-LABEL: @test_svsubwt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usubwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_u32u12__SVUint32_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usubwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CHECK-LABEL: @test_svsubwt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usubwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_u64u12__SVUint64_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usubwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// CHECK-LABEL: @test_svuqadd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svuqadd_n_s8_mu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svuqadd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s16_mu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svuqadd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s32_mu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svuqadd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s64_mu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CHECK-LABEL: @test_svuqadd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svuqadd_n_s8_zu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
// CHECK-LABEL: @test_svuqadd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s16_zu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
// CHECK-LABEL: @test_svuqadd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s32_zu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
// CHECK-LABEL: @test_svuqadd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s64_zu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
// CHECK-LABEL: @test_svuqadd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svuqadd_n_s8_xu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CHECK-LABEL: @test_svuqadd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s16_xu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CHECK-LABEL: @test_svuqadd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s32_xu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CHECK-LABEL: @test_svuqadd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s64_xu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
return IC.replaceInstUsesWith(II, Insert);
}
+static Optional<Instruction *> instCombineSVEDupX(InstCombiner &IC,
+ IntrinsicInst &II) {
+ // Replace DupX with a regular IR splat.
+ IRBuilder<> Builder(II.getContext());
+ Builder.SetInsertPoint(&II);
+ auto *RetTy = cast<ScalableVectorType>(II.getType());
+ Value *Splat =
+ Builder.CreateVectorSplat(RetTy->getElementCount(), II.getArgOperand(0));
+ Splat->takeName(&II);
+ return IC.replaceInstUsesWith(II, Splat);
+}
+
static Optional<Instruction *> instCombineSVECmpNE(InstCombiner &IC,
IntrinsicInst &II) {
LLVMContext &Ctx = II.getContext();
return None;
// Check that we have a compare of zero..
- auto *DupX = dyn_cast<IntrinsicInst>(II.getArgOperand(2));
- if (!DupX || DupX->getIntrinsicID() != Intrinsic::aarch64_sve_dup_x)
- return None;
-
- auto *DupXArg = dyn_cast<ConstantInt>(DupX->getArgOperand(0));
- if (!DupXArg || !DupXArg->isZero())
+ auto *SplatValue =
+ dyn_cast_or_null<ConstantInt>(getSplatValue(II.getArgOperand(2)));
+ if (!SplatValue || !SplatValue->isZero())
return None;
// ..against a dupq
IRBuilder<> Builder(II.getContext());
Builder.SetInsertPoint(&II);
- // Return true if a given instruction is an aarch64_sve_dup_x intrinsic call
- // with a unit splat value, false otherwise.
- auto IsUnitDupX = [](auto *I) {
- auto *IntrI = dyn_cast<IntrinsicInst>(I);
- if (!IntrI || IntrI->getIntrinsicID() != Intrinsic::aarch64_sve_dup_x)
+ // Return true if a given instruction is a unit splat value, false otherwise.
+ auto IsUnitSplat = [](auto *I) {
+ auto *SplatValue = getSplatValue(I);
+ if (!SplatValue)
return false;
-
- auto *SplatValue = IntrI->getOperand(0);
return match(SplatValue, m_FPOne()) || match(SplatValue, m_One());
};
// The OpMultiplier variable should always point to the dup (if any), so
// swap if necessary.
- if (IsUnitDup(OpMultiplicand) || IsUnitDupX(OpMultiplicand))
+ if (IsUnitDup(OpMultiplicand) || IsUnitSplat(OpMultiplicand))
std::swap(OpMultiplier, OpMultiplicand);
- if (IsUnitDupX(OpMultiplier)) {
+ if (IsUnitSplat(OpMultiplier)) {
// [f]mul pg (dupx 1) %n => %n
OpMultiplicand->takeName(&II);
return IC.replaceInstUsesWith(II, OpMultiplicand);
auto *OpIndices = II.getOperand(1);
VectorType *VTy = cast<VectorType>(II.getType());
- // Check whether OpIndices is an aarch64_sve_dup_x intrinsic call with
- // constant splat value < minimal element count of result.
- auto *DupXIntrI = dyn_cast<IntrinsicInst>(OpIndices);
- if (!DupXIntrI || DupXIntrI->getIntrinsicID() != Intrinsic::aarch64_sve_dup_x)
- return None;
-
- auto *SplatValue = dyn_cast<ConstantInt>(DupXIntrI->getOperand(0));
+ // Check whether OpIndices is a constant splat value < minimal element count
+ // of result.
+ auto *SplatValue = dyn_cast_or_null<ConstantInt>(getSplatValue(OpIndices));
if (!SplatValue ||
SplatValue->getValue().uge(VTy->getElementCount().getKnownMinValue()))
return None;
return instCombineConvertFromSVBool(IC, II);
case Intrinsic::aarch64_sve_dup:
return instCombineSVEDup(IC, II);
+ case Intrinsic::aarch64_sve_dup_x:
+ return instCombineSVEDupX(IC, II);
case Intrinsic::aarch64_sve_cmpne:
case Intrinsic::aarch64_sve_cmpne_wide:
return instCombineSVECmpNE(IC, II);
; Edge case -- make sure that the case where we're fmultiplying two dups
; together is sane.
; CHECK-LABEL: @idempotent_fmul_two_dups(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 0xH3C00)
-; CHECK-NEXT: ret <vscale x 8 x half> [[TMP1]]
+; CHECK-NEXT: ret <vscale x 8 x half> shufflevector (<vscale x 8 x half> insertelement (<vscale x 8 x half> poison, half 0xH3C00, i32 0), <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer)
;
%1 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 1.0)
%2 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 1.0)
; Non-idempotent fmuls -- we don't expect these to be optimised out.
define <vscale x 8 x half> @non_idempotent_fmul_f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a) #0 {
; CHECK-LABEL: @non_idempotent_fmul_f16(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 0xH4000)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[PG:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[PG:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> shufflevector (<vscale x 8 x half> insertelement (<vscale x 8 x half> poison, half 0xH4000, i32 0), <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 8 x half> [[TMP1]]
;
%1 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 2.0)
%2 = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
define <vscale x 4 x float> @non_idempotent_fmul_f32(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a) #0 {
; CHECK-LABEL: @non_idempotent_fmul_f32(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 2.000000e+00)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x float> [[A:%.*]], <vscale x 4 x float> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x float> [[A:%.*]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 2.000000e+00, i32 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
;
%1 = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 2.0)
%2 = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %1)
define <vscale x 2 x double> @non_idempotent_fmul_f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) #0 {
; CHECK-LABEL: @non_idempotent_fmul_f64(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 2.000000e+00)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x double> [[A:%.*]], <vscale x 2 x double> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x double> [[A:%.*]], <vscale x 2 x double> shufflevector (<vscale x 2 x double> insertelement (<vscale x 2 x double> poison, double 2.000000e+00, i32 0), <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 2 x double> [[TMP1]]
;
%1 = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 2.0)
%2 = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %1)
; Edge case -- make sure that the case where we're multiplying two dups
; together is sane.
; CHECK-LABEL: @idempotent_mul_two_dups(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
-; CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
+; CHECK-NEXT: ret <vscale x 8 x i16> shufflevector (<vscale x 8 x i16> insertelement (<vscale x 8 x i16> poison, i16 1, i32 0), <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer)
;
%1 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
%2 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
; Non-idempotent muls -- we don't expect these to be optimised out.
define <vscale x 8 x i16> @non_idempotent_mul_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) #0 {
; CHECK-LABEL: @non_idempotent_mul_i16(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 2)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[PG:%.*]], <vscale x 8 x i16> [[A:%.*]], <vscale x 8 x i16> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[PG:%.*]], <vscale x 8 x i16> [[A:%.*]], <vscale x 8 x i16> shufflevector (<vscale x 8 x i16> insertelement (<vscale x 8 x i16> poison, i16 2, i32 0), <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
;
%1 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 2)
%2 = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a, <vscale x 8 x i16> %1)
define <vscale x 4 x i32> @non_idempotent_mul_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: @non_idempotent_mul_i32(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 2)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x i32> [[A:%.*]], <vscale x 4 x i32> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x i32> [[A:%.*]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 2, i32 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
;
%1 = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 2)
%2 = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a, <vscale x 4 x i32> %1)
define <vscale x 2 x i64> @non_idempotent_mul_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) #0 {
; CHECK-LABEL: @non_idempotent_mul_i64(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 2)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x i64> [[A:%.*]], <vscale x 2 x i64> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x i64> [[A:%.*]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 2, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
;
%1 = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 2)
%2 = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a, <vscale x 2 x i64> %1)
ret <vscale x 8 x i16> %insert
}
+define <vscale x 8 x i16> @dupx_splat_convert(i16 %s) #0 {
+; CHECK-LABEL: @dupx_splat_convert(
+; CHECK-NEXT: %.splatinsert = insertelement <vscale x 8 x i16> poison, i16 %s, i32 0
+; CHECK-NEXT: %splat = shufflevector <vscale x 8 x i16> %.splatinsert, <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
+; CHECK-NEXT: ret <vscale x 8 x i16> %splat
+ %splat = tail call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 %s)
+ ret <vscale x 8 x i16> %splat
+}
+
+declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16)
+
declare <vscale x 16 x i8> @llvm.aarch64.sve.dup.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i1>, i8)
declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i1>, i16)