[LegalizeTypes][VP] Add split operand support for VP float and integer casting

Reviewed By: frasercrmck

Differential Revision: https://reviews.llvm.org/D130685

diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp b/llvm/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
index 7cafa7d..9aa9aff 100644 (file)
--- a/llvm/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
@@ -2725,6 +2725,8 @@ bool DAGTypeLegalizer::SplitVectorOperand(SDNode *N, unsigned OpNo) {
   case ISD::STRICT_UINT_TO_FP:
   case ISD::SINT_TO_FP:
   case ISD::UINT_TO_FP:
+  case ISD::VP_SINT_TO_FP:
+  case ISD::VP_UINT_TO_FP:
     if (N->getValueType(0).bitsLT(
             N->getOperand(N->isStrictFPOpcode() ? 1 : 0).getValueType()))
       Res = SplitVecOp_TruncateHelper(N);
@@ -2737,6 +2739,8 @@ bool DAGTypeLegalizer::SplitVectorOperand(SDNode *N, unsigned OpNo) {
     break;
   case ISD::FP_TO_SINT:
   case ISD::FP_TO_UINT:
+  case ISD::VP_FP_TO_SINT:
+  case ISD::VP_FP_TO_UINT:
   case ISD::STRICT_FP_TO_SINT:
   case ISD::STRICT_FP_TO_UINT:
   case ISD::STRICT_FP_EXTEND:

diff --git a/llvm/test/CodeGen/RISCV/rvv/vfptosi-vp.ll b/llvm/test/CodeGen/RISCV/rvv/vfptosi-vp.ll
index 1cff5b3..7205483 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/vfptosi-vp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfptosi-vp.ll
@@ -309,6 +309,50 @@ define <vscale x 2 x i64> @vfptosi_nxv2i64_nxv2f64_unmasked(<vscale x 2 x double
   ret <vscale x 2 x i64> %v
 }
 
+declare <vscale x 32 x i16> @llvm.vp.fptosi.nxv32i16.nxv32f32(<vscale x 32 x float>, <vscale x 32 x i1>, i32)
+
+define <vscale x 32 x i16> @vfptosi_nxv32i16_nxv32f32(<vscale x 32 x float> %va, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfptosi_nxv32i16_nxv32f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi sp, sp, -16
+; CHECK-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    slli a1, a1, 3
+; CHECK-NEXT:    sub sp, sp, a1
+; CHECK-NEXT:    vmv1r.v v24, v0
+; CHECK-NEXT:    addi a1, sp, 16
+; CHECK-NEXT:    vs8r.v v8, (a1) # Unknown-size Folded Spill
+; CHECK-NEXT:    li a2, 0
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    srli a4, a1, 2
+; CHECK-NEXT:    vsetvli a3, zero, e8, mf2, ta, mu
+; CHECK-NEXT:    slli a1, a1, 1
+; CHECK-NEXT:    sub a3, a0, a1
+; CHECK-NEXT:    vslidedown.vx v0, v0, a4
+; CHECK-NEXT:    bltu a0, a3, .LBB25_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    mv a2, a3
+; CHECK-NEXT:  .LBB25_2:
+; CHECK-NEXT:    vsetvli zero, a2, e16, m4, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v12, v16, v0.t
+; CHECK-NEXT:    bltu a0, a1, .LBB25_4
+; CHECK-NEXT:  # %bb.3:
+; CHECK-NEXT:    mv a0, a1
+; CHECK-NEXT:  .LBB25_4:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    addi a0, sp, 16
+; CHECK-NEXT:    vl8re8.v v16, (a0) # Unknown-size Folded Reload
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v8, v16, v0.t
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    slli a0, a0, 3
+; CHECK-NEXT:    add sp, sp, a0
+; CHECK-NEXT:    addi sp, sp, 16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x i16> @llvm.vp.fptosi.nxv32i16.nxv32f32(<vscale x 32 x float> %va, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x i16> %v
+}
+
 declare <vscale x 32 x i32> @llvm.vp.fptosi.nxv32i32.nxv32f32(<vscale x 32 x float>, <vscale x 32 x i1>, i32)
 
 define <vscale x 32 x i32> @vfptosi_nxv32i32_nxv32f32(<vscale x 32 x float> %va, <vscale x 32 x i1> %m, i32 zeroext %evl) {
@@ -322,16 +366,16 @@ define <vscale x 32 x i32> @vfptosi_nxv32i32_nxv32f32(<vscale x 32 x float> %va,
 ; CHECK-NEXT:    slli a1, a1, 1
 ; CHECK-NEXT:    sub a3, a0, a1
 ; CHECK-NEXT:    vslidedown.vx v0, v0, a4
-; CHECK-NEXT:    bltu a0, a3, .LBB25_2
+; CHECK-NEXT:    bltu a0, a3, .LBB26_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a2, a3
-; CHECK-NEXT:  .LBB25_2:
+; CHECK-NEXT:  .LBB26_2:
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    vfcvt.rtz.x.f.v v16, v16, v0.t
-; CHECK-NEXT:    bltu a0, a1, .LBB25_4
+; CHECK-NEXT:    bltu a0, a1, .LBB26_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mv a0, a1
-; CHECK-NEXT:  .LBB25_4:
+; CHECK-NEXT:  .LBB26_4:
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
 ; CHECK-NEXT:    vmv1r.v v0, v24
 ; CHECK-NEXT:    vfcvt.rtz.x.f.v v8, v8, v0.t
@@ -346,18 +390,18 @@ define <vscale x 32 x i32> @vfptosi_nxv32i32_nxv32f32_unmasked(<vscale x 32 x fl
 ; CHECK-NEXT:    csrr a1, vlenb
 ; CHECK-NEXT:    slli a1, a1, 1
 ; CHECK-NEXT:    mv a2, a0
-; CHECK-NEXT:    bltu a0, a1, .LBB26_2
+; CHECK-NEXT:    bltu a0, a1, .LBB27_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a2, a1
-; CHECK-NEXT:  .LBB26_2:
+; CHECK-NEXT:  .LBB27_2:
 ; CHECK-NEXT:    li a3, 0
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    sub a1, a0, a1
 ; CHECK-NEXT:    vfcvt.rtz.x.f.v v8, v8
-; CHECK-NEXT:    bltu a0, a1, .LBB26_4
+; CHECK-NEXT:    bltu a0, a1, .LBB27_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mv a3, a1
-; CHECK-NEXT:  .LBB26_4:
+; CHECK-NEXT:  .LBB27_4:
 ; CHECK-NEXT:    vsetvli zero, a3, e32, m8, ta, mu
 ; CHECK-NEXT:    vfcvt.rtz.x.f.v v16, v16
 ; CHECK-NEXT:    ret

diff --git a/llvm/test/CodeGen/RISCV/rvv/vfptoui-vp.ll b/llvm/test/CodeGen/RISCV/rvv/vfptoui-vp.ll
index 8dede1b..c3c795c 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/vfptoui-vp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfptoui-vp.ll
@@ -309,6 +309,50 @@ define <vscale x 2 x i64> @vfptoui_nxv2i64_nxv2f64_unmasked(<vscale x 2 x double
   ret <vscale x 2 x i64> %v
 }
 
+declare <vscale x 32 x i16> @llvm.vp.fptoui.nxv32i16.nxv32f32(<vscale x 32 x float>, <vscale x 32 x i1>, i32)
+
+define <vscale x 32 x i16> @vfptoui_nxv32i16_nxv32f32(<vscale x 32 x float> %va, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfptoui_nxv32i16_nxv32f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi sp, sp, -16
+; CHECK-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    slli a1, a1, 3
+; CHECK-NEXT:    sub sp, sp, a1
+; CHECK-NEXT:    vmv1r.v v24, v0
+; CHECK-NEXT:    addi a1, sp, 16
+; CHECK-NEXT:    vs8r.v v8, (a1) # Unknown-size Folded Spill
+; CHECK-NEXT:    li a2, 0
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    srli a4, a1, 2
+; CHECK-NEXT:    vsetvli a3, zero, e8, mf2, ta, mu
+; CHECK-NEXT:    slli a1, a1, 1
+; CHECK-NEXT:    sub a3, a0, a1
+; CHECK-NEXT:    vslidedown.vx v0, v0, a4
+; CHECK-NEXT:    bltu a0, a3, .LBB25_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    mv a2, a3
+; CHECK-NEXT:  .LBB25_2:
+; CHECK-NEXT:    vsetvli zero, a2, e16, m4, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v12, v16, v0.t
+; CHECK-NEXT:    bltu a0, a1, .LBB25_4
+; CHECK-NEXT:  # %bb.3:
+; CHECK-NEXT:    mv a0, a1
+; CHECK-NEXT:  .LBB25_4:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    addi a0, sp, 16
+; CHECK-NEXT:    vl8re8.v v16, (a0) # Unknown-size Folded Reload
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v8, v16, v0.t
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    slli a0, a0, 3
+; CHECK-NEXT:    add sp, sp, a0
+; CHECK-NEXT:    addi sp, sp, 16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x i16> @llvm.vp.fptoui.nxv32i16.nxv32f32(<vscale x 32 x float> %va, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x i16> %v
+}
+
 declare <vscale x 32 x i32> @llvm.vp.fptoui.nxv32i32.nxv32f32(<vscale x 32 x float>, <vscale x 32 x i1>, i32)
 
 define <vscale x 32 x i32> @vfptoui_nxv32i32_nxv32f32(<vscale x 32 x float> %va, <vscale x 32 x i1> %m, i32 zeroext %evl) {
@@ -322,16 +366,16 @@ define <vscale x 32 x i32> @vfptoui_nxv32i32_nxv32f32(<vscale x 32 x float> %va,
 ; CHECK-NEXT:    slli a1, a1, 1
 ; CHECK-NEXT:    sub a3, a0, a1
 ; CHECK-NEXT:    vslidedown.vx v0, v0, a4
-; CHECK-NEXT:    bltu a0, a3, .LBB25_2
+; CHECK-NEXT:    bltu a0, a3, .LBB26_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a2, a3
-; CHECK-NEXT:  .LBB25_2:
+; CHECK-NEXT:  .LBB26_2:
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    vfcvt.rtz.xu.f.v v16, v16, v0.t
-; CHECK-NEXT:    bltu a0, a1, .LBB25_4
+; CHECK-NEXT:    bltu a0, a1, .LBB26_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mv a0, a1
-; CHECK-NEXT:  .LBB25_4:
+; CHECK-NEXT:  .LBB26_4:
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
 ; CHECK-NEXT:    vmv1r.v v0, v24
 ; CHECK-NEXT:    vfcvt.rtz.xu.f.v v8, v8, v0.t
@@ -346,18 +390,18 @@ define <vscale x 32 x i32> @vfptoui_nxv32i32_nxv32f32_unmasked(<vscale x 32 x fl
 ; CHECK-NEXT:    csrr a1, vlenb
 ; CHECK-NEXT:    slli a1, a1, 1
 ; CHECK-NEXT:    mv a2, a0
-; CHECK-NEXT:    bltu a0, a1, .LBB26_2
+; CHECK-NEXT:    bltu a0, a1, .LBB27_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a2, a1
-; CHECK-NEXT:  .LBB26_2:
+; CHECK-NEXT:  .LBB27_2:
 ; CHECK-NEXT:    li a3, 0
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    sub a1, a0, a1
 ; CHECK-NEXT:    vfcvt.rtz.xu.f.v v8, v8
-; CHECK-NEXT:    bltu a0, a1, .LBB26_4
+; CHECK-NEXT:    bltu a0, a1, .LBB27_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mv a3, a1
-; CHECK-NEXT:  .LBB26_4:
+; CHECK-NEXT:  .LBB27_4:
 ; CHECK-NEXT:    vsetvli zero, a3, e32, m8, ta, mu
 ; CHECK-NEXT:    vfcvt.rtz.xu.f.v v16, v16
 ; CHECK-NEXT:    ret

diff --git a/llvm/test/CodeGen/RISCV/rvv/vsitofp-vp.ll b/llvm/test/CodeGen/RISCV/rvv/vsitofp-vp.ll
index 565019c..cb065ce 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/vsitofp-vp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vsitofp-vp.ll
@@ -301,6 +301,50 @@ define <vscale x 2 x double> @vsitofp_nxv2f64_nxv2i64_unmasked(<vscale x 2 x i64
   ret <vscale x 2 x double> %v
 }
 
+declare <vscale x 32 x half> @llvm.vp.sitofp.nxv32f16.nxv32i32(<vscale x 32 x i32>, <vscale x 32 x i1>, i32)
+
+define <vscale x 32 x half> @vsitofp_nxv32f16_nxv32i32(<vscale x 32 x i32> %va, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vsitofp_nxv32f16_nxv32i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi sp, sp, -16
+; CHECK-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    slli a1, a1, 3
+; CHECK-NEXT:    sub sp, sp, a1
+; CHECK-NEXT:    vmv1r.v v24, v0
+; CHECK-NEXT:    addi a1, sp, 16
+; CHECK-NEXT:    vs8r.v v8, (a1) # Unknown-size Folded Spill
+; CHECK-NEXT:    li a2, 0
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    srli a4, a1, 2
+; CHECK-NEXT:    vsetvli a3, zero, e8, mf2, ta, mu
+; CHECK-NEXT:    slli a1, a1, 1
+; CHECK-NEXT:    sub a3, a0, a1
+; CHECK-NEXT:    vslidedown.vx v0, v0, a4
+; CHECK-NEXT:    bltu a0, a3, .LBB25_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    mv a2, a3
+; CHECK-NEXT:  .LBB25_2:
+; CHECK-NEXT:    vsetvli zero, a2, e16, m4, ta, mu
+; CHECK-NEXT:    vfncvt.f.x.w v12, v16, v0.t
+; CHECK-NEXT:    bltu a0, a1, .LBB25_4
+; CHECK-NEXT:  # %bb.3:
+; CHECK-NEXT:    mv a0, a1
+; CHECK-NEXT:  .LBB25_4:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    addi a0, sp, 16
+; CHECK-NEXT:    vl8re8.v v16, (a0) # Unknown-size Folded Reload
+; CHECK-NEXT:    vfncvt.f.x.w v8, v16, v0.t
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    slli a0, a0, 3
+; CHECK-NEXT:    add sp, sp, a0
+; CHECK-NEXT:    addi sp, sp, 16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x half> @llvm.vp.sitofp.nxv32f16.nxv32i32(<vscale x 32 x i32> %va, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
 declare <vscale x 32 x float> @llvm.vp.sitofp.nxv32f32.nxv32i32(<vscale x 32 x i32>, <vscale x 32 x i1>, i32)
 
 define <vscale x 32 x float> @vsitofp_nxv32f32_nxv32i32(<vscale x 32 x i32> %va, <vscale x 32 x i1> %m, i32 zeroext %evl) {
@@ -314,16 +358,16 @@ define <vscale x 32 x float> @vsitofp_nxv32f32_nxv32i32(<vscale x 32 x i32> %va,
 ; CHECK-NEXT:    slli a1, a1, 1
 ; CHECK-NEXT:    sub a3, a0, a1
 ; CHECK-NEXT:    vslidedown.vx v0, v0, a4
-; CHECK-NEXT:    bltu a0, a3, .LBB25_2
+; CHECK-NEXT:    bltu a0, a3, .LBB26_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a2, a3
-; CHECK-NEXT:  .LBB25_2:
+; CHECK-NEXT:  .LBB26_2:
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    vfcvt.f.x.v v16, v16, v0.t
-; CHECK-NEXT:    bltu a0, a1, .LBB25_4
+; CHECK-NEXT:    bltu a0, a1, .LBB26_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mv a0, a1
-; CHECK-NEXT:  .LBB25_4:
+; CHECK-NEXT:  .LBB26_4:
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
 ; CHECK-NEXT:    vmv1r.v v0, v24
 ; CHECK-NEXT:    vfcvt.f.x.v v8, v8, v0.t
@@ -338,18 +382,18 @@ define <vscale x 32 x float> @vsitofp_nxv32f32_nxv32i32_unmasked(<vscale x 32 x
 ; CHECK-NEXT:    csrr a1, vlenb
 ; CHECK-NEXT:    slli a1, a1, 1
 ; CHECK-NEXT:    mv a2, a0
-; CHECK-NEXT:    bltu a0, a1, .LBB26_2
+; CHECK-NEXT:    bltu a0, a1, .LBB27_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a2, a1
-; CHECK-NEXT:  .LBB26_2:
+; CHECK-NEXT:  .LBB27_2:
 ; CHECK-NEXT:    li a3, 0
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    sub a1, a0, a1
 ; CHECK-NEXT:    vfcvt.f.x.v v8, v8
-; CHECK-NEXT:    bltu a0, a1, .LBB26_4
+; CHECK-NEXT:    bltu a0, a1, .LBB27_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mv a3, a1
-; CHECK-NEXT:  .LBB26_4:
+; CHECK-NEXT:  .LBB27_4:
 ; CHECK-NEXT:    vsetvli zero, a3, e32, m8, ta, mu
 ; CHECK-NEXT:    vfcvt.f.x.v v16, v16
 ; CHECK-NEXT:    ret

diff --git a/llvm/test/CodeGen/RISCV/rvv/vuitofp-vp.ll b/llvm/test/CodeGen/RISCV/rvv/vuitofp-vp.ll
index da002b7..85ddb37 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/vuitofp-vp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vuitofp-vp.ll
@@ -301,6 +301,50 @@ define <vscale x 2 x double> @vuitofp_nxv2f64_nxv2i64_unmasked(<vscale x 2 x i64
   ret <vscale x 2 x double> %v
 }
 
+declare <vscale x 32 x half> @llvm.vp.uitofp.nxv32f16.nxv32i32(<vscale x 32 x i32>, <vscale x 32 x i1>, i32)
+
+define <vscale x 32 x half> @vuitofp_nxv32f16_nxv32i32(<vscale x 32 x i32> %va, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vuitofp_nxv32f16_nxv32i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi sp, sp, -16
+; CHECK-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    slli a1, a1, 3
+; CHECK-NEXT:    sub sp, sp, a1
+; CHECK-NEXT:    vmv1r.v v24, v0
+; CHECK-NEXT:    addi a1, sp, 16
+; CHECK-NEXT:    vs8r.v v8, (a1) # Unknown-size Folded Spill
+; CHECK-NEXT:    li a2, 0
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    srli a4, a1, 2
+; CHECK-NEXT:    vsetvli a3, zero, e8, mf2, ta, mu
+; CHECK-NEXT:    slli a1, a1, 1
+; CHECK-NEXT:    sub a3, a0, a1
+; CHECK-NEXT:    vslidedown.vx v0, v0, a4
+; CHECK-NEXT:    bltu a0, a3, .LBB25_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    mv a2, a3
+; CHECK-NEXT:  .LBB25_2:
+; CHECK-NEXT:    vsetvli zero, a2, e16, m4, ta, mu
+; CHECK-NEXT:    vfncvt.f.xu.w v12, v16, v0.t
+; CHECK-NEXT:    bltu a0, a1, .LBB25_4
+; CHECK-NEXT:  # %bb.3:
+; CHECK-NEXT:    mv a0, a1
+; CHECK-NEXT:  .LBB25_4:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    addi a0, sp, 16
+; CHECK-NEXT:    vl8re8.v v16, (a0) # Unknown-size Folded Reload
+; CHECK-NEXT:    vfncvt.f.xu.w v8, v16, v0.t
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    slli a0, a0, 3
+; CHECK-NEXT:    add sp, sp, a0
+; CHECK-NEXT:    addi sp, sp, 16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x half> @llvm.vp.uitofp.nxv32f16.nxv32i32(<vscale x 32 x i32> %va, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
 declare <vscale x 32 x float> @llvm.vp.uitofp.nxv32f32.nxv32i32(<vscale x 32 x i32>, <vscale x 32 x i1>, i32)
 
 define <vscale x 32 x float> @vuitofp_nxv32f32_nxv32i32(<vscale x 32 x i32> %va, <vscale x 32 x i1> %m, i32 zeroext %evl) {
@@ -314,16 +358,16 @@ define <vscale x 32 x float> @vuitofp_nxv32f32_nxv32i32(<vscale x 32 x i32> %va,
 ; CHECK-NEXT:    slli a1, a1, 1
 ; CHECK-NEXT:    sub a3, a0, a1
 ; CHECK-NEXT:    vslidedown.vx v0, v0, a4
-; CHECK-NEXT:    bltu a0, a3, .LBB25_2
+; CHECK-NEXT:    bltu a0, a3, .LBB26_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a2, a3
-; CHECK-NEXT:  .LBB25_2:
+; CHECK-NEXT:  .LBB26_2:
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    vfcvt.f.xu.v v16, v16, v0.t
-; CHECK-NEXT:    bltu a0, a1, .LBB25_4
+; CHECK-NEXT:    bltu a0, a1, .LBB26_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mv a0, a1
-; CHECK-NEXT:  .LBB25_4:
+; CHECK-NEXT:  .LBB26_4:
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
 ; CHECK-NEXT:    vmv1r.v v0, v24
 ; CHECK-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
@@ -338,18 +382,18 @@ define <vscale x 32 x float> @vuitofp_nxv32f32_nxv32i32_unmasked(<vscale x 32 x
 ; CHECK-NEXT:    csrr a1, vlenb
 ; CHECK-NEXT:    slli a1, a1, 1
 ; CHECK-NEXT:    mv a2, a0
-; CHECK-NEXT:    bltu a0, a1, .LBB26_2
+; CHECK-NEXT:    bltu a0, a1, .LBB27_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a2, a1
-; CHECK-NEXT:  .LBB26_2:
+; CHECK-NEXT:  .LBB27_2:
 ; CHECK-NEXT:    li a3, 0
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    sub a1, a0, a1
 ; CHECK-NEXT:    vfcvt.f.xu.v v8, v8
-; CHECK-NEXT:    bltu a0, a1, .LBB26_4
+; CHECK-NEXT:    bltu a0, a1, .LBB27_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mv a3, a1
-; CHECK-NEXT:  .LBB26_4:
+; CHECK-NEXT:  .LBB27_4:
 ; CHECK-NEXT:    vsetvli zero, a3, e32, m8, ta, mu
 ; CHECK-NEXT:    vfcvt.f.xu.v v16, v16
 ; CHECK-NEXT:    ret