[riscv] add test coverage for fractional lmul w/fixed length vectorization

diff --git a/llvm/test/CodeGen/RISCV/rvv/sink-splat-operands.ll b/llvm/test/CodeGen/RISCV/rvv/sink-splat-operands.ll
index 15716a8..0a3772e 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/sink-splat-operands.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/sink-splat-operands.ll
@@ -4281,3 +4281,255 @@ vector.body:                                      ; preds = %vector.body, %entry
 for.cond.cleanup:                                 ; preds = %vector.body
   ret void
 }
+
+define void @sink_splat_mul_lmulmf2(i64* nocapture %a, i32 signext %x) {
+; CHECK-LABEL: sink_splat_mul_lmulmf2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    li a2, 1024
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:  .LBB81_1: # %vector.body
+; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    vle32.v v8, (a0)
+; CHECK-NEXT:    vmul.vx v8, v8, a1
+; CHECK-NEXT:    vse32.v v8, (a0)
+; CHECK-NEXT:    addi a2, a2, -4
+; CHECK-NEXT:    addi a0, a0, 32
+; CHECK-NEXT:    bnez a2, .LBB81_1
+; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
+; CHECK-NEXT:    ret
+entry:
+  %broadcast.splatinsert = insertelement <2 x i32> poison, i32 %x, i64 0
+  %broadcast.splat = shufflevector <2 x i32> %broadcast.splatinsert, <2 x i32> poison, <2 x i32> zeroinitializer
+  br label %vector.body
+
+vector.body:                                      ; preds = %vector.body, %entry
+  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
+  %0 = getelementptr inbounds i64, i64* %a, i64 %index
+  %1 = bitcast i64* %0 to <2 x i32>*
+  %wide.load = load <2 x i32>, <2 x i32>* %1, align 8
+  %2 = mul <2 x i32> %wide.load, %broadcast.splat
+  %3 = bitcast i64* %0 to <2 x i32>*
+  store <2 x i32> %2, <2 x i32>* %3, align 8
+  %index.next = add nuw i64 %index, 4
+  %4 = icmp eq i64 %index.next, 1024
+  br i1 %4, label %for.cond.cleanup, label %vector.body
+
+for.cond.cleanup:                                 ; preds = %vector.body
+  ret void
+}
+
+define void @sink_splat_add_lmulmf2(i64* nocapture %a, i32 signext %x) {
+; CHECK-LABEL: sink_splat_add_lmulmf2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    li a2, 1024
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:  .LBB82_1: # %vector.body
+; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    vle32.v v8, (a0)
+; CHECK-NEXT:    vadd.vx v8, v8, a1
+; CHECK-NEXT:    vse32.v v8, (a0)
+; CHECK-NEXT:    addi a2, a2, -4
+; CHECK-NEXT:    addi a0, a0, 32
+; CHECK-NEXT:    bnez a2, .LBB82_1
+; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
+; CHECK-NEXT:    ret
+entry:
+  %broadcast.splatinsert = insertelement <2 x i32> poison, i32 %x, i64 0
+  %broadcast.splat = shufflevector <2 x i32> %broadcast.splatinsert, <2 x i32> poison, <2 x i32> zeroinitializer
+  br label %vector.body
+
+vector.body:                                      ; preds = %vector.body, %entry
+  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
+  %0 = getelementptr inbounds i64, i64* %a, i64 %index
+  %1 = bitcast i64* %0 to <2 x i32>*
+  %wide.load = load <2 x i32>, <2 x i32>* %1, align 8
+  %2 = add <2 x i32> %wide.load, %broadcast.splat
+  %3 = bitcast i64* %0 to <2 x i32>*
+  store <2 x i32> %2, <2 x i32>* %3, align 8
+  %index.next = add nuw i64 %index, 4
+  %4 = icmp eq i64 %index.next, 1024
+  br i1 %4, label %for.cond.cleanup, label %vector.body
+
+for.cond.cleanup:                                 ; preds = %vector.body
+  ret void
+}
+
+define void @sink_splat_sub_lmulmf2(i64* nocapture %a, i32 signext %x) {
+; CHECK-LABEL: sink_splat_sub_lmulmf2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    li a2, 1024
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:  .LBB83_1: # %vector.body
+; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    vle32.v v8, (a0)
+; CHECK-NEXT:    vsub.vx v8, v8, a1
+; CHECK-NEXT:    vse32.v v8, (a0)
+; CHECK-NEXT:    addi a2, a2, -4
+; CHECK-NEXT:    addi a0, a0, 32
+; CHECK-NEXT:    bnez a2, .LBB83_1
+; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
+; CHECK-NEXT:    ret
+entry:
+  %broadcast.splatinsert = insertelement <2 x i32> poison, i32 %x, i64 0
+  %broadcast.splat = shufflevector <2 x i32> %broadcast.splatinsert, <2 x i32> poison, <2 x i32> zeroinitializer
+  br label %vector.body
+
+vector.body:                                      ; preds = %vector.body, %entry
+  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
+  %0 = getelementptr inbounds i64, i64* %a, i64 %index
+  %1 = bitcast i64* %0 to <2 x i32>*
+  %wide.load = load <2 x i32>, <2 x i32>* %1, align 8
+  %2 = sub <2 x i32> %wide.load, %broadcast.splat
+  %3 = bitcast i64* %0 to <2 x i32>*
+  store <2 x i32> %2, <2 x i32>* %3, align 8
+  %index.next = add nuw i64 %index, 4
+  %4 = icmp eq i64 %index.next, 1024
+  br i1 %4, label %for.cond.cleanup, label %vector.body
+
+for.cond.cleanup:                                 ; preds = %vector.body
+  ret void
+}
+
+define void @sink_splat_rsub_lmulmf2(i64* nocapture %a, i32 signext %x) {
+; CHECK-LABEL: sink_splat_rsub_lmulmf2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    li a2, 1024
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:  .LBB84_1: # %vector.body
+; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    vle32.v v8, (a0)
+; CHECK-NEXT:    vrsub.vx v8, v8, a1
+; CHECK-NEXT:    vse32.v v8, (a0)
+; CHECK-NEXT:    addi a2, a2, -4
+; CHECK-NEXT:    addi a0, a0, 32
+; CHECK-NEXT:    bnez a2, .LBB84_1
+; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
+; CHECK-NEXT:    ret
+entry:
+  %broadcast.splatinsert = insertelement <2 x i32> poison, i32 %x, i64 0
+  %broadcast.splat = shufflevector <2 x i32> %broadcast.splatinsert, <2 x i32> poison, <2 x i32> zeroinitializer
+  br label %vector.body
+
+vector.body:                                      ; preds = %vector.body, %entry
+  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
+  %0 = getelementptr inbounds i64, i64* %a, i64 %index
+  %1 = bitcast i64* %0 to <2 x i32>*
+  %wide.load = load <2 x i32>, <2 x i32>* %1, align 8
+  %2 = sub <2 x i32> %broadcast.splat, %wide.load
+  %3 = bitcast i64* %0 to <2 x i32>*
+  store <2 x i32> %2, <2 x i32>* %3, align 8
+  %index.next = add nuw i64 %index, 4
+  %4 = icmp eq i64 %index.next, 1024
+  br i1 %4, label %for.cond.cleanup, label %vector.body
+
+for.cond.cleanup:                                 ; preds = %vector.body
+  ret void
+}
+
+define void @sink_splat_and_lmulmf2(i64* nocapture %a, i32 signext %x) {
+; CHECK-LABEL: sink_splat_and_lmulmf2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    li a2, 1024
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:  .LBB85_1: # %vector.body
+; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    vle32.v v8, (a0)
+; CHECK-NEXT:    vand.vx v8, v8, a1
+; CHECK-NEXT:    vse32.v v8, (a0)
+; CHECK-NEXT:    addi a2, a2, -4
+; CHECK-NEXT:    addi a0, a0, 32
+; CHECK-NEXT:    bnez a2, .LBB85_1
+; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
+; CHECK-NEXT:    ret
+entry:
+  %broadcast.splatinsert = insertelement <2 x i32> poison, i32 %x, i64 0
+  %broadcast.splat = shufflevector <2 x i32> %broadcast.splatinsert, <2 x i32> poison, <2 x i32> zeroinitializer
+  br label %vector.body
+
+vector.body:                                      ; preds = %vector.body, %entry
+  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
+  %0 = getelementptr inbounds i64, i64* %a, i64 %index
+  %1 = bitcast i64* %0 to <2 x i32>*
+  %wide.load = load <2 x i32>, <2 x i32>* %1, align 8
+  %2 = and <2 x i32> %wide.load, %broadcast.splat
+  %3 = bitcast i64* %0 to <2 x i32>*
+  store <2 x i32> %2, <2 x i32>* %3, align 8
+  %index.next = add nuw i64 %index, 4
+  %4 = icmp eq i64 %index.next, 1024
+  br i1 %4, label %for.cond.cleanup, label %vector.body
+
+for.cond.cleanup:                                 ; preds = %vector.body
+  ret void
+}
+
+define void @sink_splat_or_lmulmf2(i64* nocapture %a, i32 signext %x) {
+; CHECK-LABEL: sink_splat_or_lmulmf2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    li a2, 1024
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:  .LBB86_1: # %vector.body
+; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    vle32.v v8, (a0)
+; CHECK-NEXT:    vor.vx v8, v8, a1
+; CHECK-NEXT:    vse32.v v8, (a0)
+; CHECK-NEXT:    addi a2, a2, -4
+; CHECK-NEXT:    addi a0, a0, 32
+; CHECK-NEXT:    bnez a2, .LBB86_1
+; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
+; CHECK-NEXT:    ret
+entry:
+  %broadcast.splatinsert = insertelement <2 x i32> poison, i32 %x, i64 0
+  %broadcast.splat = shufflevector <2 x i32> %broadcast.splatinsert, <2 x i32> poison, <2 x i32> zeroinitializer
+  br label %vector.body
+
+vector.body:                                      ; preds = %vector.body, %entry
+  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
+  %0 = getelementptr inbounds i64, i64* %a, i64 %index
+  %1 = bitcast i64* %0 to <2 x i32>*
+  %wide.load = load <2 x i32>, <2 x i32>* %1, align 8
+  %2 = or <2 x i32> %wide.load, %broadcast.splat
+  %3 = bitcast i64* %0 to <2 x i32>*
+  store <2 x i32> %2, <2 x i32>* %3, align 8
+  %index.next = add nuw i64 %index, 4
+  %4 = icmp eq i64 %index.next, 1024
+  br i1 %4, label %for.cond.cleanup, label %vector.body
+
+for.cond.cleanup:                                 ; preds = %vector.body
+  ret void
+}
+
+define void @sink_splat_xor_lmulmf2(i64* nocapture %a, i32 signext %x) {
+; CHECK-LABEL: sink_splat_xor_lmulmf2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    li a2, 1024
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:  .LBB87_1: # %vector.body
+; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    vle32.v v8, (a0)
+; CHECK-NEXT:    vxor.vx v8, v8, a1
+; CHECK-NEXT:    vse32.v v8, (a0)
+; CHECK-NEXT:    addi a2, a2, -4
+; CHECK-NEXT:    addi a0, a0, 32
+; CHECK-NEXT:    bnez a2, .LBB87_1
+; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
+; CHECK-NEXT:    ret
+entry:
+  %broadcast.splatinsert = insertelement <2 x i32> poison, i32 %x, i64 0
+  %broadcast.splat = shufflevector <2 x i32> %broadcast.splatinsert, <2 x i32> poison, <2 x i32> zeroinitializer
+  br label %vector.body
+
+vector.body:                                      ; preds = %vector.body, %entry
+  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
+  %0 = getelementptr inbounds i64, i64* %a, i64 %index
+  %1 = bitcast i64* %0 to <2 x i32>*
+  %wide.load = load <2 x i32>, <2 x i32>* %1, align 8
+  %2 = xor <2 x i32> %wide.load, %broadcast.splat
+  %3 = bitcast i64* %0 to <2 x i32>*
+  store <2 x i32> %2, <2 x i32>* %3, align 8
+  %index.next = add nuw i64 %index, 4
+  %4 = icmp eq i64 %index.next, 1024
+  br i1 %4, label %for.cond.cleanup, label %vector.body
+
+for.cond.cleanup:                                 ; preds = %vector.body
+  ret void
+}