[RISCV][COST] Add half type support for the cost model of sqrt/fabs

[Miss_Grape]

Summary

1. Refactor for costs of sqrt/fabs
2. Add half type support for the cost model of sqrt/fabs

[reames]

Please update the review description to describe actual change. In this case, I think you're adding half support.

llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp

681

Unrelated change. Feel free to simply land a change which swaps the order here, but please remove it from this diff.

[Miss_Grape]

ping

[Miss_Grape]

ping

[Miss_Grape]

ping

[craig.topper]

llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp

704

ST->hasVInstructionF16() is only true if hasVInstructions() is true so this || doesn't do anything.

[Miss_Grape]

adress the comment

[Miss_Grape]

ping

[Miss_Grape]

ping

[craig.topper]

LGTM

llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp

static const CostTblEntry VectorIntrinsicCostTable[]{
     {Intrinsic::roundeven, MVT::v2f64, 9},
     {Intrinsic::roundeven, MVT::v4f64, 9},
     {Intrinsic::roundeven, MVT::v8f64, 9},
     {Intrinsic::roundeven, MVT::v16f64, 9},
     {Intrinsic::roundeven, MVT::nxv1f64, 9},
     {Intrinsic::roundeven, MVT::nxv2f64, 9},
     {Intrinsic::roundeven, MVT::nxv4f64, 9},
     {Intrinsic::roundeven, MVT::nxv8f64, 9},
-    {Intrinsic::fabs, MVT::v2f32, 1},
-    {Intrinsic::fabs, MVT::v4f32, 1},
-    {Intrinsic::fabs, MVT::v8f32, 1},
-    {Intrinsic::fabs, MVT::v16f32, 1},
-    {Intrinsic::fabs, MVT::nxv1f32, 1},
-    {Intrinsic::fabs, MVT::nxv2f32, 1},
-    {Intrinsic::fabs, MVT::nxv4f32, 1},
-    {Intrinsic::fabs, MVT::nxv8f32, 1},
-    {Intrinsic::fabs, MVT::nxv16f32, 1},
-    {Intrinsic::fabs, MVT::v2f64, 1},
-    {Intrinsic::fabs, MVT::v4f64, 1},
-    {Intrinsic::fabs, MVT::v8f64, 1},
-    {Intrinsic::fabs, MVT::v16f64, 1},
-    {Intrinsic::fabs, MVT::nxv1f64, 1},
-    {Intrinsic::fabs, MVT::nxv2f64, 1},
-    {Intrinsic::fabs, MVT::nxv4f64, 1},
-    {Intrinsic::fabs, MVT::nxv8f64, 1},
-    {Intrinsic::sqrt, MVT::v2f32, 1},
-    {Intrinsic::sqrt, MVT::v4f32, 1},
-    {Intrinsic::sqrt, MVT::v8f32, 1},
-    {Intrinsic::sqrt, MVT::v16f32, 1},
-    {Intrinsic::sqrt, MVT::nxv1f32, 1},
-    {Intrinsic::sqrt, MVT::nxv2f32, 1},
-    {Intrinsic::sqrt, MVT::nxv4f32, 1},
-    {Intrinsic::sqrt, MVT::nxv8f32, 1},
-    {Intrinsic::sqrt, MVT::nxv16f32, 1},
-    {Intrinsic::sqrt, MVT::v2f64, 1},
-    {Intrinsic::sqrt, MVT::v4f64, 1},
-    {Intrinsic::sqrt, MVT::v8f64, 1},
-    {Intrinsic::sqrt, MVT::v16f64, 1},
-    {Intrinsic::sqrt, MVT::nxv1f64, 1},
-    {Intrinsic::sqrt, MVT::nxv2f64, 1},
-    {Intrinsic::sqrt, MVT::nxv4f64, 1},
-    {Intrinsic::sqrt, MVT::nxv8f64, 1},
     {Intrinsic::bswap, MVT::v2i16, 3},
     {Intrinsic::bswap, MVT::v4i16, 3},
     {Intrinsic::bswap, MVT::v8i16, 3},
     {Intrinsic::bswap, MVT::v16i16, 3},
     {Intrinsic::bswap, MVT::nxv1i16, 3},
     {Intrinsic::bswap, MVT::nxv2i16, 3},
     {Intrinsic::bswap, MVT::nxv4i16, 3},
     {Intrinsic::bswap, MVT::nxv8i16, 3},
case Intrinsic::roundeven: {
     break;
   }
   case Intrinsic::umin:
   case Intrinsic::umax:
   case Intrinsic::smin:
   case Intrinsic::smax: {
     auto LT = getTypeLegalizationCost(RetTy);
     if ((ST->hasVInstructions() && LT.second.isVector()) ||
         (LT.second.isScalarInteger() && ST->hasStdExtZbb()))
       return LT.first;
     break;
   }
+  case Intrinsic::fabs:
+  case Intrinsic::sqrt: {
+    auto LT = getTypeLegalizationCost(RetTy);
+    if (ST->hasVInstructions() && LT.second.isVector())
+      return LT.first;
+    break;
+  }
   case Intrinsic::sadd_sat:
   case Intrinsic::ssub_sat:
   case Intrinsic::uadd_sat:
   case Intrinsic::usub_sat: {
     auto LT = getTypeLegalizationCost(RetTy);
     if (ST->hasVInstructions() && LT.second.isVector())
       return LT.first;
     break;
   }
   // TODO: add more intrinsic
   case Intrinsic::experimental_stepvector: {
     unsigned Cost = 1; // vid
     auto LT = getTypeLegalizationCost(RetTy);
     return Cost + (LT.first - 1);
   }
   case Intrinsic::vp_rint: {
     // RISC-V target uses at least 5 instructions to lower rounding intrinsics.
     unsigned Cost = 5;
     auto LT = getTypeLegalizationCost(RetTy);
     if (TLI->isOperationCustom(ISD::VP_FRINT, LT.second))
       return Cost * LT.first;

llvm/test/Analysis/CostModel/RISCV/fp-min-max-abs.ll

 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
-; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -S -mtriple=riscv64 -mattr=+v,+f,+d -riscv-v-vector-bits-min=-1 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -S -mtriple=riscv64 -mattr=+v,+f,+d,+zfh,+experimental-zvfh -riscv-v-vector-bits-min=-1 | FileCheck %s
 
 define void @fabs() {
 ; CHECK-LABEL: 'fabs'
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = call float @llvm.fabs.f32(float undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %2 = call <2 x float> @llvm.fabs.v2f32(<2 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %3 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %4 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %5 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %6 = call <vscale x 1 x float> @llvm.fabs.nxv1f32(<vscale x 1 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %7 = call <vscale x 2 x float> @llvm.fabs.nxv2f32(<vscale x 2 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %8 = call <vscale x 4 x float> @llvm.fabs.nxv4f32(<vscale x 4 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %9 = call <vscale x 8 x float> @llvm.fabs.nxv8f32(<vscale x 8 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %10 = call <vscale x 16 x float> @llvm.fabs.nxv16f32(<vscale x 16 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %11 = call double @llvm.fabs.f64(double undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %12 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %13 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %14 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %15 = call <16 x double> @llvm.fabs.v16f64(<16 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %16 = call <vscale x 1 x double> @llvm.fabs.nxv1f64(<vscale x 1 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %17 = call <vscale x 2 x double> @llvm.fabs.nxv2f64(<vscale x 2 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %18 = call <vscale x 4 x double> @llvm.fabs.nxv4f64(<vscale x 4 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %19 = call <vscale x 8 x double> @llvm.fabs.nxv8f64(<vscale x 8 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %2 = call half @llvm.fabs.f16(half undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %3 = call <2 x half> @llvm.fabs.v2f16(<2 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %4 = call <4 x half> @llvm.fabs.v4f16(<4 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %5 = call <8 x half> @llvm.fabs.v8f16(<8 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %6 = call <16 x half> @llvm.fabs.v16f16(<16 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %7 = call <2 x float> @llvm.fabs.v2f32(<2 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %8 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %9 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %10 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %11 = call <vscale x 2 x half> @llvm.fabs.nxv2f16(<vscale x 2 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %12 = call <vscale x 4 x half> @llvm.fabs.nxv4f16(<vscale x 4 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %13 = call <vscale x 8 x half> @llvm.fabs.nxv8f16(<vscale x 8 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %14 = call <vscale x 16 x half> @llvm.fabs.nxv16f16(<vscale x 16 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %15 = call <vscale x 1 x float> @llvm.fabs.nxv1f32(<vscale x 1 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %16 = call <vscale x 2 x float> @llvm.fabs.nxv2f32(<vscale x 2 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %17 = call <vscale x 4 x float> @llvm.fabs.nxv4f32(<vscale x 4 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %18 = call <vscale x 8 x float> @llvm.fabs.nxv8f32(<vscale x 8 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %19 = call <vscale x 16 x float> @llvm.fabs.nxv16f32(<vscale x 16 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %20 = call double @llvm.fabs.f64(double undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %21 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %22 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %23 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %24 = call <16 x double> @llvm.fabs.v16f64(<16 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %25 = call <vscale x 1 x double> @llvm.fabs.nxv1f64(<vscale x 1 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %26 = call <vscale x 2 x double> @llvm.fabs.nxv2f64(<vscale x 2 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %27 = call <vscale x 4 x double> @llvm.fabs.nxv4f64(<vscale x 4 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %28 = call <vscale x 8 x double> @llvm.fabs.nxv8f64(<vscale x 8 x double> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   call float @llvm.fabs.f32(float undef)
+  call half @llvm.fabs.f16(half undef)
+  call <2 x half> @llvm.fabs.v2f16(<2 x half> undef)
+  call <4 x half> @llvm.fabs.v4f16(<4 x half> undef)
+  call <8 x half> @llvm.fabs.v8f16(<8 x half> undef)
+  call <16 x half> @llvm.fabs.v16f16(<16 x half> undef)
   call <2 x float> @llvm.fabs.v2f32(<2 x float> undef)
   call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
   call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
   call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+  call <vscale x 2 x half> @llvm.fabs.nvx2f16(<vscale x 2 x half> undef)
+  call <vscale x 4 x half> @llvm.fabs.nvx4f16(<vscale x 4 x half> undef)
+  call <vscale x 8 x half> @llvm.fabs.nvx8f16(<vscale x 8 x half> undef)
+  call <vscale x 16 x half> @llvm.fabs.nvx16f16(<vscale x 16 x half> undef)
   call <vscale x 1 x float> @llvm.fabs.nvx1f32(<vscale x 1 x float> undef)
   call <vscale x 2 x float> @llvm.fabs.nvx2f32(<vscale x 2 x float> undef)
   call <vscale x 4 x float> @llvm.fabs.nvx4f32(<vscale x 4 x float> undef)
   call <vscale x 8 x float> @llvm.fabs.nvx8f32(<vscale x 8 x float> undef)
   call <vscale x 16 x float> @llvm.fabs.nvx16f32(<vscale x 16 x float> undef)
   call double @llvm.fabs.f64(double undef)
   call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
   call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
;
   call <vscale x 1 x double> @llvm.copysign.nvx1f64(<vscale x 1 x double> undef, <vscale x 1 x double> undef)
   call <vscale x 2 x double> @llvm.copysign.nvx2f64(<vscale x 2 x double> undef, <vscale x 2 x double> undef)
   call <vscale x 4 x double> @llvm.copysign.nvx4f64(<vscale x 4 x double> undef, <vscale x 4 x double> undef)
   call <vscale x 8 x double> @llvm.copysign.nvx8f64(<vscale x 8 x double> undef, <vscale x 8 x double> undef)
   ret void
 }
 
 declare float @llvm.fabs.f32(float)
+declare half @llvm.fabs.f16(half)
+declare <2 x half> @llvm.fabs.v2f16(<2 x half>)
+declare <4 x half> @llvm.fabs.v4f16(<4 x half>)
+declare <8 x half> @llvm.fabs.v8f16(<8 x half>)
+declare <16 x half> @llvm.fabs.v16f16(<16 x half>)
 declare <2 x float> @llvm.fabs.v2f32(<2 x float>)
 declare <4 x float> @llvm.fabs.v4f32(<4 x float>)
 declare <8 x float> @llvm.fabs.v8f32(<8 x float>)
 declare <16 x float> @llvm.fabs.v16f32(<16 x float>)
+declare <vscale x 2 x half> @llvm.fabs.nvx2f16(<vscale x 2 x half>)
+declare <vscale x 4 x half> @llvm.fabs.nvx4f16(<vscale x 4 x half>)
+declare <vscale x 8 x half> @llvm.fabs.nvx8f16(<vscale x 8 x half>)
+declare <vscale x 16 x half> @llvm.fabs.nvx16f16(<vscale x 16 x half>)
 declare <vscale x 1 x float> @llvm.fabs.nvx1f32(<vscale x 1 x float>)
 declare <vscale x 2 x float> @llvm.fabs.nvx2f32(<vscale x 2 x float>)
 declare <vscale x 4 x float> @llvm.fabs.nvx4f32(<vscale x 4 x float>)
 declare <vscale x 8 x float> @llvm.fabs.nvx8f32(<vscale x 8 x float>)
 declare <vscale x 16 x float> @llvm.fabs.nvx16f32(<vscale x 16 x float>)
 declare double @llvm.fabs.f64(double)
 declare <2 x double> @llvm.fabs.v2f64(<2 x double>)
 declare <4 x double> @llvm.fabs.v4f64(<4 x double>)

llvm/test/Analysis/CostModel/RISCV/fp-sqrt-pow.ll

 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
-; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -S -mtriple=riscv64 -mattr=+v,+f,+d -riscv-v-vector-bits-min=-1 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -S -mtriple=riscv64 -mattr=+v,+f,+d,+zfh,+experimental-zvfh -riscv-v-vector-bits-min=-1 | FileCheck %s
 
 define void @sqrt() {
 ; CHECK-LABEL: 'sqrt'
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = call float @llvm.sqrt.f32(float undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %2 = call <2 x float> @llvm.sqrt.v2f32(<2 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %3 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %4 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %5 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %6 = call <vscale x 1 x float> @llvm.sqrt.nxv1f32(<vscale x 1 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %7 = call <vscale x 2 x float> @llvm.sqrt.nxv2f32(<vscale x 2 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %8 = call <vscale x 4 x float> @llvm.sqrt.nxv4f32(<vscale x 4 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %9 = call <vscale x 8 x float> @llvm.sqrt.nxv8f32(<vscale x 8 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %10 = call <vscale x 16 x float> @llvm.sqrt.nxv16f32(<vscale x 16 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %11 = call double @llvm.sqrt.f64(double undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %12 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %13 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %14 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %15 = call <16 x double> @llvm.sqrt.v16f64(<16 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %16 = call <vscale x 1 x double> @llvm.sqrt.nxv1f64(<vscale x 1 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %17 = call <vscale x 2 x double> @llvm.sqrt.nxv2f64(<vscale x 2 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %18 = call <vscale x 4 x double> @llvm.sqrt.nxv4f64(<vscale x 4 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %19 = call <vscale x 8 x double> @llvm.sqrt.nxv8f64(<vscale x 8 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %2 = call half @llvm.sqrt.f16(half undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %3 = call <2 x half> @llvm.sqrt.v2f16(<2 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %4 = call <4 x half> @llvm.sqrt.v4f16(<4 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %5 = call <8 x half> @llvm.sqrt.v8f16(<8 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %6 = call <16 x half> @llvm.sqrt.v16f16(<16 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %7 = call <2 x float> @llvm.sqrt.v2f32(<2 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %8 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %9 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %10 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %11 = call <vscale x 2 x half> @llvm.sqrt.nxv2f16(<vscale x 2 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %12 = call <vscale x 4 x half> @llvm.sqrt.nxv4f16(<vscale x 4 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %13 = call <vscale x 8 x half> @llvm.sqrt.nxv8f16(<vscale x 8 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %14 = call <vscale x 16 x half> @llvm.sqrt.nxv16f16(<vscale x 16 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %15 = call <vscale x 1 x float> @llvm.sqrt.nxv1f32(<vscale x 1 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %16 = call <vscale x 2 x float> @llvm.sqrt.nxv2f32(<vscale x 2 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %17 = call <vscale x 4 x float> @llvm.sqrt.nxv4f32(<vscale x 4 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %18 = call <vscale x 8 x float> @llvm.sqrt.nxv8f32(<vscale x 8 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %19 = call <vscale x 16 x float> @llvm.sqrt.nxv16f32(<vscale x 16 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %20 = call double @llvm.sqrt.f64(double undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %21 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %22 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %23 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %24 = call <16 x double> @llvm.sqrt.v16f64(<16 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %25 = call <vscale x 1 x double> @llvm.sqrt.nxv1f64(<vscale x 1 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %26 = call <vscale x 2 x double> @llvm.sqrt.nxv2f64(<vscale x 2 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %27 = call <vscale x 4 x double> @llvm.sqrt.nxv4f64(<vscale x 4 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %28 = call <vscale x 8 x double> @llvm.sqrt.nxv8f64(<vscale x 8 x double> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   call float @llvm.sqrt.f32(float undef)
+  call half @llvm.sqrt.f16(half undef)
+  call <2 x half> @llvm.sqrt.v2f16(<2 x half> undef)
+  call <4 x half> @llvm.sqrt.v4f16(<4 x half> undef)
+  call <8 x half> @llvm.sqrt.v8f16(<8 x half> undef)
+  call <16 x half> @llvm.sqrt.v16f16(<16 x half> undef)
   call <2 x float> @llvm.sqrt.v2f32(<2 x float> undef)
   call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
   call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
   call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+  call <vscale x 2 x half> @llvm.sqrt.nvx2f16(<vscale x 2 x half> undef)
+  call <vscale x 4 x half> @llvm.sqrt.nvx4f16(<vscale x 4 x half> undef)
+  call <vscale x 8 x half> @llvm.sqrt.nvx8f16(<vscale x 8 x half> undef)
+  call <vscale x 16 x half> @llvm.sqrt.nvx16f16(<vscale x 16 x half> undef)
   call <vscale x 1 x float> @llvm.sqrt.nvx1f32(<vscale x 1 x float> undef)
   call <vscale x 2 x float> @llvm.sqrt.nvx2f32(<vscale x 2 x float> undef)
   call <vscale x 4 x float> @llvm.sqrt.nvx4f32(<vscale x 4 x float> undef)
   call <vscale x 8 x float> @llvm.sqrt.nvx8f32(<vscale x 8 x float> undef)
   call <vscale x 16 x float> @llvm.sqrt.nvx16f32(<vscale x 16 x float> undef)
   call double @llvm.sqrt.f64(double undef)
   call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
   call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
   call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
   call <16 x double> @llvm.sqrt.v16f64(<16 x double> undef)
   call <vscale x 1 x double> @llvm.sqrt.nvx1f64(<vscale x 1 x double> undef)
   call <vscale x 2 x double> @llvm.sqrt.nvx2f64(<vscale x 2 x double> undef)
   call <vscale x 4 x double> @llvm.sqrt.nvx4f64(<vscale x 4 x double> undef)
   call <vscale x 8 x double> @llvm.sqrt.nvx8f64(<vscale x 8 x double> undef)
   ret void
 }
 
+declare half @llvm.sqrt.f16(half)
 declare float @llvm.sqrt.f32(float)
+declare <2 x half> @llvm.sqrt.v2f16(<2 x half>)
+declare <4 x half> @llvm.sqrt.v4f16(<4 x half>)
+declare <8 x half> @llvm.sqrt.v8f16(<8 x half>)
+declare <16 x half> @llvm.sqrt.v16f16(<16 x half>)
 declare <2 x float> @llvm.sqrt.v2f32(<2 x float>)
 declare <4 x float> @llvm.sqrt.v4f32(<4 x float>)
 declare <8 x float> @llvm.sqrt.v8f32(<8 x float>)
 declare <16 x float> @llvm.sqrt.v16f32(<16 x float>)
+declare <vscale x 2 x half> @llvm.sqrt.nvx2f16(<vscale x 2 x half>)
+declare <vscale x 4 x half> @llvm.sqrt.nvx4f16(<vscale x 4 x half>)
+declare <vscale x 8 x half> @llvm.sqrt.nvx8f16(<vscale x 8 x half>)
+declare <vscale x 16 x half> @llvm.sqrt.nvx16f16(<vscale x 16 x half>)
 declare <vscale x 1 x float> @llvm.sqrt.nvx1f32(<vscale x 1 x float>)
 declare <vscale x 2 x float> @llvm.sqrt.nvx2f32(<vscale x 2 x float>)
 declare <vscale x 4 x float> @llvm.sqrt.nvx4f32(<vscale x 4 x float>)
 declare <vscale x 8 x float> @llvm.sqrt.nvx8f32(<vscale x 8 x float>)
 declare <vscale x 16 x float> @llvm.sqrt.nvx16f32(<vscale x 16 x float>)
 declare double @llvm.sqrt.f64(double)
 declare <2 x double> @llvm.sqrt.v2f64(<2 x double>)
 declare <4 x double> @llvm.sqrt.v4f64(<4 x double>)