From: David Green Date: Thu, 19 May 2022 18:53:21 +0000 (+0100) Subject: [ARM] Cost modelling for scalar fptoi_sat X-Git-Tag: upstream/15.0.7~7205 X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=80aab0312acea2231b9aec4ba2ceb5951ae8016a;p=platform%2Fupstream%2Fllvm.git [ARM] Cost modelling for scalar fptoi_sat Similar to D124357, this adds some cost modelling for fptoi_sat for Arm targets. Where VFP2 is available (and FP64/FP16 for the relevant types), the operations are legal as the Arm instructions naturally saturate. Otherwise they will need an extra smin/smax clamp, similar to AArch64. Differential Revision: https://reviews.llvm.org/D125665 --- diff --git a/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp b/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp index 9d376ad..8742163 100644 --- a/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp +++ b/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp @@ -1765,6 +1765,39 @@ ARMTTIImpl::getIntrinsicInstrCost(const IntrinsicCostAttributes &ICA, return LT.first * ST->getMVEVectorCostFactor(CostKind); break; } + case Intrinsic::fptosi_sat: + case Intrinsic::fptoui_sat: { + if (ICA.getArgTypes().empty()) + break; + bool IsSigned = ICA.getID() == Intrinsic::fptosi_sat; + auto LT = TLI->getTypeLegalizationCost(DL, ICA.getArgTypes()[0]); + EVT MTy = TLI->getValueType(DL, ICA.getReturnType()); + // Check for the legal types, with the corect subtarget features. + if ((ST->hasVFP2Base() && LT.second == MVT::f32 && MTy == MVT::i32) || + (ST->hasFP64() && LT.second == MVT::f64 && MTy == MVT::i32) || + (ST->hasFullFP16() && LT.second == MVT::f16 && MTy == MVT::i32)) + return LT.first; + + // Otherwise we use a legal convert followed by a min+max + if (((ST->hasVFP2Base() && LT.second == MVT::f32) || + (ST->hasFP64() && LT.second == MVT::f64) || + (ST->hasFullFP16() && LT.second == MVT::f16)) && + LT.second.getScalarSizeInBits() >= MTy.getScalarSizeInBits()) { + Type *LegalTy = Type::getIntNTy(ICA.getReturnType()->getContext(), + LT.second.getScalarSizeInBits()); + InstructionCost Cost = 1; + IntrinsicCostAttributes Attrs1(IsSigned ? Intrinsic::smin + : Intrinsic::umin, + LegalTy, {LegalTy, LegalTy}); + Cost += getIntrinsicInstrCost(Attrs1, CostKind); + IntrinsicCostAttributes Attrs2(IsSigned ? Intrinsic::smax + : Intrinsic::umax, + LegalTy, {LegalTy, LegalTy}); + Cost += getIntrinsicInstrCost(Attrs2, CostKind); + return LT.first * Cost; + } + break; + } } return BaseT::getIntrinsicInstrCost(ICA, CostKind); diff --git a/llvm/test/Analysis/CostModel/ARM/fptoi_sat.ll b/llvm/test/Analysis/CostModel/ARM/fptoi_sat.ll index 72d4c28..8a7784c 100644 --- a/llvm/test/Analysis/CostModel/ARM/fptoi_sat.ll +++ b/llvm/test/Analysis/CostModel/ARM/fptoi_sat.ll @@ -1,6 +1,6 @@ ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py ; RUN: opt -passes='print' 2>&1 -disable-output -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve < %s | FileCheck %s --check-prefix=CHECK-MVE -; RUN: opt -passes='print' 2>&1 -disable-output -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve.fp < %s | FileCheck %s --check-prefix=CHECK-MVEFP +; RUN: opt -passes='print' 2>&1 -disable-output -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve.fp,+fp64 < %s | FileCheck %s --check-prefix=CHECK-MVEFP define void @casts() { ; CHECK-MVE-LABEL: 'casts' @@ -108,25 +108,25 @@ define void @casts() { ; ; CHECK-MVEFP-LABEL: 'casts' ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f32s1 = call i1 @llvm.fptosi.sat.i1.f32(float undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %f32u1 = call i1 @llvm.fptoui.sat.i1.f32(float undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f32u1 = call i1 @llvm.fptoui.sat.i1.f32(float undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f32s8 = call i8 @llvm.fptosi.sat.i8.f32(float undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %f32u8 = call i8 @llvm.fptoui.sat.i8.f32(float undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f32u8 = call i8 @llvm.fptoui.sat.i8.f32(float undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f32s16 = call i16 @llvm.fptosi.sat.i16.f32(float undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %f32u16 = call i16 @llvm.fptoui.sat.i16.f32(float undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f32s32 = call i32 @llvm.fptosi.sat.i32.f32(float undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %f32u32 = call i32 @llvm.fptoui.sat.i32.f32(float undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f32u16 = call i16 @llvm.fptoui.sat.i16.f32(float undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f32s32 = call i32 @llvm.fptosi.sat.i32.f32(float undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f32u32 = call i32 @llvm.fptoui.sat.i32.f32(float undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %f32s64 = call i64 @llvm.fptosi.sat.i64.f32(float undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %f32u64 = call i64 @llvm.fptoui.sat.i64.f32(float undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %f64s1 = call i1 @llvm.fptosi.sat.i1.f64(double undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %f64u1 = call i1 @llvm.fptoui.sat.i1.f64(double undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %f64s8 = call i8 @llvm.fptosi.sat.i8.f64(double undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %f64u8 = call i8 @llvm.fptoui.sat.i8.f64(double undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %f64s16 = call i16 @llvm.fptosi.sat.i16.f64(double undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %f64u16 = call i16 @llvm.fptoui.sat.i16.f64(double undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %f64s32 = call i32 @llvm.fptosi.sat.i32.f64(double undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %f64u32 = call i32 @llvm.fptoui.sat.i32.f64(double undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %f64s64 = call i64 @llvm.fptosi.sat.i64.f64(double undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %f64u64 = call i64 @llvm.fptoui.sat.i64.f64(double undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %f64s1 = call i1 @llvm.fptosi.sat.i1.f64(double undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %f64u1 = call i1 @llvm.fptoui.sat.i1.f64(double undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %f64s8 = call i8 @llvm.fptosi.sat.i8.f64(double undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %f64u8 = call i8 @llvm.fptoui.sat.i8.f64(double undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %f64s16 = call i16 @llvm.fptosi.sat.i16.f64(double undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %f64u16 = call i16 @llvm.fptoui.sat.i16.f64(double undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f64s32 = call i32 @llvm.fptosi.sat.i32.f64(double undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f64u32 = call i32 @llvm.fptoui.sat.i32.f64(double undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %f64s64 = call i64 @llvm.fptosi.sat.i64.f64(double undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %f64u64 = call i64 @llvm.fptoui.sat.i64.f64(double undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %v2f32s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f32(<2 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %v2f32u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f32(<2 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %v2f32s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f32(<2 x float> undef) @@ -137,16 +137,16 @@ define void @casts() { ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %v2f32u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f32(<2 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 76 for instruction: %v2f32s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f32(<2 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %v2f32u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f32(<2 x float> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 116 for instruction: %v2f64s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f64(<2 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %v2f64u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f64(<2 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 98 for instruction: %v2f64s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f64(<2 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %v2f64u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f64(<2 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 98 for instruction: %v2f64s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f64(<2 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %v2f64u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f64(<2 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 98 for instruction: %v2f64s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f64(<2 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %v2f64u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f64(<2 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 130 for instruction: %v2f64s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f64(<2 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 120 for instruction: %v2f64u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f64(<2 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %v2f64s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f64(<2 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %v2f64u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f64(<2 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %v2f64s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f64(<2 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %v2f64u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f64(<2 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %v2f64s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f64(<2 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %v2f64u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f64(<2 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %v2f64s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f64(<2 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %v2f64u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f64(<2 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 94 for instruction: %v2f64s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f64(<2 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %v2f64u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f64(<2 x double> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %v4f32s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f32(<4 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v4f32u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f32(<4 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %v4f32s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f32(<4 x float> undef) @@ -157,16 +157,16 @@ define void @casts() { ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v4f32u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f32(<4 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 284 for instruction: %v4f32s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f32(<4 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 278 for instruction: %v4f32u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f32(<4 x float> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 306 for instruction: %v4f64s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f64(<4 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 250 for instruction: %v4f64u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f64(<4 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 268 for instruction: %v4f64s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f64(<4 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 250 for instruction: %v4f64u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f64(<4 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 268 for instruction: %v4f64s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f64(<4 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 250 for instruction: %v4f64u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f64(<4 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 268 for instruction: %v4f64s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f64(<4 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 250 for instruction: %v4f64u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f64(<4 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 396 for instruction: %v4f64s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f64(<4 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 376 for instruction: %v4f64u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f64(<4 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 234 for instruction: %v4f64s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f64(<4 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 178 for instruction: %v4f64u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f64(<4 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 196 for instruction: %v4f64s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f64(<4 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 178 for instruction: %v4f64u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f64(<4 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 196 for instruction: %v4f64s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f64(<4 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 178 for instruction: %v4f64u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f64(<4 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 196 for instruction: %v4f64s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f64(<4 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 178 for instruction: %v4f64u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f64(<4 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 324 for instruction: %v4f64s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f64(<4 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 304 for instruction: %v4f64u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f64(<4 x double> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 134 for instruction: %v8f32s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f32(<8 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %v8f32u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f32(<8 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %v8f32s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f32(<8 x float> undef) @@ -177,16 +177,16 @@ define void @casts() { ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %v8f32u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f32(<8 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1148 for instruction: %v8f32s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f32(<8 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1104 for instruction: %v8f32u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f32(<8 x float> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 906 for instruction: %v8f64s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f64(<8 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 794 for instruction: %v8f64u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f64(<8 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 828 for instruction: %v8f64s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f64(<8 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 794 for instruction: %v8f64u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f64(<8 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 828 for instruction: %v8f64s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f64(<8 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 794 for instruction: %v8f64u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f64(<8 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 828 for instruction: %v8f64s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f64(<8 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 792 for instruction: %v8f64u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f64(<8 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1336 for instruction: %v8f64s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f64(<8 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1296 for instruction: %v8f64u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f64(<8 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 762 for instruction: %v8f64s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f64(<8 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 650 for instruction: %v8f64u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f64(<8 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 684 for instruction: %v8f64s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f64(<8 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 650 for instruction: %v8f64u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f64(<8 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 684 for instruction: %v8f64s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f64(<8 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 650 for instruction: %v8f64u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f64(<8 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 684 for instruction: %v8f64s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f64(<8 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 648 for instruction: %v8f64u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f64(<8 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1192 for instruction: %v8f64s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f64(<8 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1152 for instruction: %v8f64u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f64(<8 x double> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 290 for instruction: %v16f32s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f32(<16 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 58 for instruction: %v16f32u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f32(<16 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 132 for instruction: %v16f32s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f32(<16 x float> undef) @@ -197,16 +197,16 @@ define void @casts() { ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %v16f32u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f32(<16 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4488 for instruction: %v16f32s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f32(<16 x float> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4400 for instruction: %v16f32u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f32(<16 x float> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2986 for instruction: %v16f64s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f64(<16 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2762 for instruction: %v16f64u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f64(<16 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2828 for instruction: %v16f64s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f64(<16 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2762 for instruction: %v16f64u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f64(<16 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2828 for instruction: %v16f64s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f64(<16 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2760 for instruction: %v16f64u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f64(<16 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2824 for instruction: %v16f64s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f64(<16 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2752 for instruction: %v16f64u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f64(<16 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4848 for instruction: %v16f64s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f64(<16 x double> undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4768 for instruction: %v16f64u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f64(<16 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2698 for instruction: %v16f64s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f64(<16 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2474 for instruction: %v16f64u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f64(<16 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2540 for instruction: %v16f64s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f64(<16 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2474 for instruction: %v16f64u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f64(<16 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2540 for instruction: %v16f64s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f64(<16 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2472 for instruction: %v16f64u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f64(<16 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2536 for instruction: %v16f64s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f64(<16 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2464 for instruction: %v16f64u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f64(<16 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4560 for instruction: %v16f64s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f64(<16 x double> undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4480 for instruction: %v16f64u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f64(<16 x double> undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void ; %f32s1 = call i1 @llvm.fptosi.sat.i1.f32(float undef) @@ -378,13 +378,13 @@ define void @fp16() { ; ; CHECK-MVEFP-LABEL: 'fp16' ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f16s1 = call i1 @llvm.fptosi.sat.i1.f16(half undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %f16u1 = call i1 @llvm.fptoui.sat.i1.f16(half undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f16u1 = call i1 @llvm.fptoui.sat.i1.f16(half undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f16s8 = call i8 @llvm.fptosi.sat.i8.f16(half undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %f16u8 = call i8 @llvm.fptoui.sat.i8.f16(half undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f16u8 = call i8 @llvm.fptoui.sat.i8.f16(half undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f16s16 = call i16 @llvm.fptosi.sat.i16.f16(half undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %f16u16 = call i16 @llvm.fptoui.sat.i16.f16(half undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f16s32 = call i32 @llvm.fptosi.sat.i32.f16(half undef) -; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %f16u32 = call i32 @llvm.fptoui.sat.i32.f16(half undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %f16u16 = call i16 @llvm.fptoui.sat.i16.f16(half undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16s32 = call i32 @llvm.fptosi.sat.i32.f16(half undef) +; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16u32 = call i32 @llvm.fptoui.sat.i32.f16(half undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %f16s64 = call i64 @llvm.fptosi.sat.i64.f16(half undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %f16u64 = call i64 @llvm.fptoui.sat.i64.f16(half undef) ; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %v2f16s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f16(<2 x half> undef)