mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-12-14 15:19:33 +00:00
AVX-512: Updated cost of FP/SINT/UINT conversion operations
I checked and updated the cost of AVX-512 conversion operations. Added cost of conversion operations in DQ mode. Conversion of illegal types that requires vector split is not calculated right now (like for other X86 targets). Differential Revision: http://reviews.llvm.org/D15074 llvm-svn: 254494
This commit is contained in:
Elena Demikhovsky
parent
d6d08d5567
commit
32e8d60483
@ -528,17 +528,31 @@ int X86TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst, Type *Src) {
|
||||
int ISD = TLI->InstructionOpcodeToISD(Opcode);
|
||||
assert(ISD && "Invalid opcode");
|
||||
|
||||
static const TypeConversionCostTblEntry AVX512ConversionTbl[] = {
|
||||
static const TypeConversionCostTblEntry AVX512DQConversionTbl[] = {
|
||||
{ ISD::UINT_TO_FP, MVT::v2f64, MVT::v2i64, 1 },
|
||||
{ ISD::UINT_TO_FP, MVT::v4f64, MVT::v4i64, 1 },
|
||||
{ ISD::UINT_TO_FP, MVT::v8f64, MVT::v8i64, 1 },
|
||||
{ ISD::UINT_TO_FP, MVT::v2f32, MVT::v2i64, 1 },
|
||||
{ ISD::UINT_TO_FP, MVT::v4f32, MVT::v4i64, 1 },
|
||||
{ ISD::UINT_TO_FP, MVT::v8f32, MVT::v8i64, 1 },
|
||||
|
||||
{ ISD::FP_TO_UINT, MVT::v2i64, MVT::v2f64, 1 },
|
||||
{ ISD::FP_TO_UINT, MVT::v4i64, MVT::v4f64, 1 },
|
||||
{ ISD::FP_TO_UINT, MVT::v8i64, MVT::v8f64, 1 },
|
||||
{ ISD::FP_TO_UINT, MVT::v2i64, MVT::v2f32, 1 },
|
||||
{ ISD::FP_TO_UINT, MVT::v4i64, MVT::v4f32, 1 },
|
||||
{ ISD::FP_TO_UINT, MVT::v8i64, MVT::v8f32, 1 },
|
||||
};
|
||||
|
||||
static const TypeConversionCostTblEntry AVX512FConversionTbl[] = {
|
||||
{ ISD::FP_EXTEND, MVT::v8f64, MVT::v8f32, 1 },
|
||||
{ ISD::FP_EXTEND, MVT::v8f64, MVT::v16f32, 3 },
|
||||
{ ISD::FP_ROUND, MVT::v8f32, MVT::v8f64, 1 },
|
||||
{ ISD::FP_ROUND, MVT::v16f32, MVT::v8f64, 3 },
|
||||
|
||||
{ ISD::TRUNCATE, MVT::v16i8, MVT::v16i32, 1 },
|
||||
{ ISD::TRUNCATE, MVT::v16i16, MVT::v16i32, 1 },
|
||||
{ ISD::TRUNCATE, MVT::v8i16, MVT::v8i64, 1 },
|
||||
{ ISD::TRUNCATE, MVT::v8i32, MVT::v8i64, 1 },
|
||||
{ ISD::TRUNCATE, MVT::v16i32, MVT::v8i64, 4 },
|
||||
|
||||
// v16i1 -> v16i32 - load + broadcast
|
||||
{ ISD::SIGN_EXTEND, MVT::v16i32, MVT::v16i1, 2 },
|
||||
@ -548,16 +562,46 @@ int X86TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst, Type *Src) {
|
||||
{ ISD::ZERO_EXTEND, MVT::v16i32, MVT::v16i8, 1 },
|
||||
{ ISD::SIGN_EXTEND, MVT::v16i32, MVT::v16i16, 1 },
|
||||
{ ISD::ZERO_EXTEND, MVT::v16i32, MVT::v16i16, 1 },
|
||||
{ ISD::SIGN_EXTEND, MVT::v8i64, MVT::v16i32, 3 },
|
||||
{ ISD::ZERO_EXTEND, MVT::v8i64, MVT::v16i32, 3 },
|
||||
{ ISD::SIGN_EXTEND, MVT::v8i64, MVT::v8i32, 1 },
|
||||
{ ISD::ZERO_EXTEND, MVT::v8i64, MVT::v8i32, 1 },
|
||||
{ ISD::ZERO_EXTEND, MVT::v8i64, MVT::v8i16, 1 },
|
||||
{ ISD::SIGN_EXTEND, MVT::v8i64, MVT::v8i16, 1 },
|
||||
|
||||
{ ISD::SINT_TO_FP, MVT::v16f32, MVT::v16i1, 3 },
|
||||
{ ISD::SINT_TO_FP, MVT::v16f32, MVT::v16i8, 2 },
|
||||
{ ISD::SINT_TO_FP, MVT::v16f32, MVT::v16i16, 2 },
|
||||
{ ISD::SINT_TO_FP, MVT::v16f32, MVT::v16i32, 1 },
|
||||
{ ISD::SINT_TO_FP, MVT::v8f64, MVT::v8i1, 4 },
|
||||
{ ISD::SINT_TO_FP, MVT::v8f64, MVT::v8i8, 2 },
|
||||
{ ISD::SINT_TO_FP, MVT::v8f64, MVT::v8i16, 2 },
|
||||
{ ISD::SINT_TO_FP, MVT::v8f64, MVT::v8i32, 1 },
|
||||
|
||||
{ ISD::UINT_TO_FP, MVT::v16f32, MVT::v16i1, 3 },
|
||||
{ ISD::UINT_TO_FP, MVT::v16f32, MVT::v16i8, 2 },
|
||||
{ ISD::UINT_TO_FP, MVT::v16f32, MVT::v16i16, 2 },
|
||||
{ ISD::UINT_TO_FP, MVT::v16f32, MVT::v16i32, 1 },
|
||||
{ ISD::UINT_TO_FP, MVT::v8f32, MVT::v8i32, 1 },
|
||||
{ ISD::UINT_TO_FP, MVT::v4f32, MVT::v4i32, 1 },
|
||||
{ ISD::UINT_TO_FP, MVT::v8f64, MVT::v8i1, 4 },
|
||||
{ ISD::UINT_TO_FP, MVT::v8f64, MVT::v8i16, 2 },
|
||||
{ ISD::UINT_TO_FP, MVT::v8f64, MVT::v8i32, 1 },
|
||||
{ ISD::UINT_TO_FP, MVT::v8f64, MVT::v8i8, 2 },
|
||||
{ ISD::UINT_TO_FP, MVT::v8f32, MVT::v8i8, 2 },
|
||||
{ ISD::UINT_TO_FP, MVT::v8f32, MVT::v8i16, 2 },
|
||||
{ ISD::UINT_TO_FP, MVT::v4f64, MVT::v4i8, 2 },
|
||||
{ ISD::UINT_TO_FP, MVT::v4f64, MVT::v4i16, 2 },
|
||||
{ ISD::UINT_TO_FP, MVT::v4f64, MVT::v4i32, 1 },
|
||||
{ ISD::UINT_TO_FP, MVT::v2f64, MVT::v2i8, 2 },
|
||||
{ ISD::UINT_TO_FP, MVT::v2f64, MVT::v2i16, 5 },
|
||||
{ ISD::UINT_TO_FP, MVT::v2f32, MVT::v2i32, 2 },
|
||||
{ ISD::UINT_TO_FP, MVT::v2f64, MVT::v2i64, 5 },
|
||||
{ ISD::UINT_TO_FP, MVT::v4f64, MVT::v4i64, 12 },
|
||||
{ ISD::UINT_TO_FP, MVT::v8f64, MVT::v8i64, 26 },
|
||||
|
||||
{ ISD::FP_TO_UINT, MVT::v2i32, MVT::v2f32, 1 },
|
||||
{ ISD::FP_TO_UINT, MVT::v4i32, MVT::v4f32, 1 },
|
||||
{ ISD::FP_TO_UINT, MVT::v8i32, MVT::v8f32, 1 },
|
||||
{ ISD::FP_TO_UINT, MVT::v16i32, MVT::v16f32, 1 },
|
||||
};
|
||||
|
||||
static const TypeConversionCostTblEntry AVX2ConversionTbl[] = {
|
||||
@ -693,12 +737,6 @@ int X86TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst, Type *Src) {
|
||||
return LTSrc.first * Entry->Cost;
|
||||
}
|
||||
|
||||
if (ST->hasAVX512()) {
|
||||
if (const auto *Entry = ConvertCostTableLookup(AVX512ConversionTbl, ISD,
|
||||
LTDest.second, LTSrc.second))
|
||||
return Entry->Cost;
|
||||
}
|
||||
|
||||
EVT SrcTy = TLI->getValueType(DL, Src);
|
||||
EVT DstTy = TLI->getValueType(DL, Dst);
|
||||
|
||||
@ -706,6 +744,18 @@ int X86TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst, Type *Src) {
|
||||
if (!SrcTy.isSimple() || !DstTy.isSimple())
|
||||
return BaseT::getCastInstrCost(Opcode, Dst, Src);
|
||||
|
||||
if (ST->hasDQI())
|
||||
if (const auto *Entry = ConvertCostTableLookup(AVX512DQConversionTbl, ISD,
|
||||
DstTy.getSimpleVT(),
|
||||
SrcTy.getSimpleVT()))
|
||||
return Entry->Cost;
|
||||
|
||||
if (ST->hasAVX512())
|
||||
if (const auto *Entry = ConvertCostTableLookup(AVX512FConversionTbl, ISD,
|
||||
DstTy.getSimpleVT(),
|
||||
SrcTy.getSimpleVT()))
|
||||
return Entry->Cost;
|
||||
|
||||
if (ST->hasAVX2()) {
|
||||
if (const auto *Entry = ConvertCostTableLookup(AVX2ConversionTbl, ISD,
|
||||
DstTy.getSimpleVT(),
|
||||
|
||||
@ -84,11 +84,11 @@ define i32 @zext_sext(<8 x i1> %in) {
|
||||
;CHECK-AVX: cost of 4 {{.*}} zext
|
||||
%D = zext <4 x i32> undef to <4 x i64>
|
||||
|
||||
;CHECK-AVX512: cost of 3 {{.*}} %D1 = zext
|
||||
%D1 = zext <16 x i32> undef to <16 x i64>
|
||||
;CHECK-AVX512: cost of 1 {{.*}} %D1 = zext
|
||||
%D1 = zext <8 x i32> undef to <8 x i64>
|
||||
|
||||
;CHECK-AVX512: cost of 3 {{.*}} %D2 = sext
|
||||
%D2 = sext <16 x i32> undef to <16 x i64>
|
||||
;CHECK-AVX512: cost of 1 {{.*}} %D2 = sext
|
||||
%D2 = sext <8 x i32> undef to <8 x i64>
|
||||
|
||||
;CHECK-AVX512: cost of 1 {{.*}} %D3 = zext
|
||||
%D3 = zext <16 x i16> undef to <16 x i32>
|
||||
@ -118,9 +118,11 @@ define i32 @zext_sext(<8 x i1> %in) {
|
||||
;CHECK_AVX512: cost of 1 {{.*}} G = trunc
|
||||
%G = trunc <8 x i64> undef to <8 x i32>
|
||||
|
||||
;CHECK-AVX512: cost of 4 {{.*}} %G1 = trunc
|
||||
%G1 = trunc <16 x i64> undef to <16 x i32>
|
||||
;CHECK-AVX512: cost of 1 {{.*}} %G1 = trunc
|
||||
%G1 = trunc <16 x i32> undef to <16 x i16>
|
||||
|
||||
;CHECK-AVX512: cost of 1 {{.*}} %G2 = trunc
|
||||
%G2 = trunc <16 x i32> undef to <16 x i8>
|
||||
ret i32 undef
|
||||
}
|
||||
|
||||
@ -207,38 +209,40 @@ define void @uitofp4(<4 x i1> %a, <4 x i8> %b, <4 x i16> %c, <4 x i32> %d) {
|
||||
; CHECK: cost of 2 {{.*}} uitofp
|
||||
%C2 = uitofp <4 x i16> %c to <4 x double>
|
||||
|
||||
; CHECK: cost of 6 {{.*}} uitofp
|
||||
; CHECK-AVX2: cost of 6 {{.*}} uitofp
|
||||
%D1 = uitofp <4 x i32> %d to <4 x float>
|
||||
; CHECK: cost of 6 {{.*}} uitofp
|
||||
; CHECK-AVX2: cost of 6 {{.*}} uitofp
|
||||
%D2 = uitofp <4 x i32> %d to <4 x double>
|
||||
ret void
|
||||
}
|
||||
|
||||
define void @uitofp8(<8 x i1> %a, <8 x i8> %b, <8 x i16> %c, <8 x i32> %d) {
|
||||
; CHECK-LABEL: for function 'uitofp8'
|
||||
; CHECK: cost of 6 {{.*}} uitofp
|
||||
; CHECK-AVX2: cost of 6 {{.*}} uitofp
|
||||
%A1 = uitofp <8 x i1> %a to <8 x float>
|
||||
|
||||
; CHECK: cost of 5 {{.*}} uitofp
|
||||
; CHECK-AVX2: cost of 5 {{.*}} uitofp
|
||||
; CHECK-AVX512: cost of 2 {{.*}} uitofp
|
||||
%B1 = uitofp <8 x i8> %b to <8 x float>
|
||||
|
||||
; CHECK: cost of 5 {{.*}} uitofp
|
||||
; CHECK-AVX2: cost of 5 {{.*}} uitofp
|
||||
; CHECK-AVX512: cost of 2 {{.*}} uitofp
|
||||
%C1 = uitofp <8 x i16> %c to <8 x float>
|
||||
|
||||
; CHECK-AVX2: cost of 8 {{.*}} uitofp
|
||||
; CHECK-AVX512: cost of 8 {{.*}} uitofp
|
||||
; CHECK-AVX512: cost of 1 {{.*}} uitofp
|
||||
; CHECK-AVX: cost of 9 {{.*}} uitofp
|
||||
%D1 = uitofp <8 x i32> %d to <8 x float>
|
||||
ret void
|
||||
}
|
||||
|
||||
define void @fp_conv(<8 x float> %a, <16 x float>%b) {
|
||||
define void @fp_conv(<8 x float> %a, <16 x float>%b, <4 x float> %c) {
|
||||
;CHECK-LABEL: for function 'fp_conv'
|
||||
; CHECK-AVX512: cost of 1 {{.*}} fpext
|
||||
%A1 = fpext <8 x float> %a to <8 x double>
|
||||
|
||||
; CHECK-AVX512: cost of 3 {{.*}} fpext
|
||||
%A2 = fpext <16 x float> %b to <16 x double>
|
||||
; CHECK-AVX512: cost of 1 {{.*}} fpext
|
||||
%A2 = fpext <4 x float> %c to <4 x double>
|
||||
|
||||
; CHECK-AVX2: cost of 3 {{.*}} %A3 = fpext
|
||||
; CHECK-AVX512: cost of 1 {{.*}} %A3 = fpext
|
||||
@ -248,7 +252,7 @@ define void @fp_conv(<8 x float> %a, <16 x float>%b) {
|
||||
; CHECK-AVX512: cost of 1 {{.*}} %A4 = fptrunc
|
||||
%A4 = fptrunc <8 x double> undef to <8 x float>
|
||||
|
||||
; CHECK-AVX512: cost of 3 {{.*}} %A5 = fptrunc
|
||||
%A5 = fptrunc <16 x double> undef to <16 x float>
|
||||
; CHECK-AVX512: cost of 1 {{.*}} %A5 = fptrunc
|
||||
%A5 = fptrunc <4 x double> undef to <4 x float>
|
||||
ret void
|
||||
}
|
||||
|
||||
@ -4,656 +4,656 @@
|
||||
; RUN: opt -mtriple=x86_64-apple-darwin -mattr=+avx512f -cost-model -analyze < %s | FileCheck --check-prefix=AVX512F %s
|
||||
|
||||
define <2 x double> @sitofpv2i8v2double(<2 x i8> %a) {
|
||||
; SSE2: sitofpv2i8v2double
|
||||
; SSE2-LABEL: sitofpv2i8v2double
|
||||
; SSE2: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv2i8v2double
|
||||
; AVX1-LABEL: sitofpv2i8v2double
|
||||
; AVX1: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv2i8v2double
|
||||
; AVX2-LABEL: sitofpv2i8v2double
|
||||
; AVX2: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv2i8v2double
|
||||
; AVX512F-LABEL: sitofpv2i8v2double
|
||||
; AVX512F: cost of 4 {{.*}} sitofp
|
||||
%1 = sitofp <2 x i8> %a to <2 x double>
|
||||
ret <2 x double> %1
|
||||
}
|
||||
|
||||
define <4 x double> @sitofpv4i8v4double(<4 x i8> %a) {
|
||||
; SSE2: sitofpv4i8v4double
|
||||
; SSE2-LABEL: sitofpv4i8v4double
|
||||
; SSE2: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv4i8v4double
|
||||
; AVX1-LABEL: sitofpv4i8v4double
|
||||
; AVX1: cost of 3 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv4i8v4double
|
||||
; AVX2-LABEL: sitofpv4i8v4double
|
||||
; AVX2: cost of 3 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv4i8v4double
|
||||
; AVX512F-LABEL: sitofpv4i8v4double
|
||||
; AVX512F: cost of 3 {{.*}} sitofp
|
||||
%1 = sitofp <4 x i8> %a to <4 x double>
|
||||
ret <4 x double> %1
|
||||
}
|
||||
|
||||
define <8 x double> @sitofpv8i8v8double(<8 x i8> %a) {
|
||||
; SSE2: sitofpv8i8v8double
|
||||
; SSE2-LABEL: sitofpv8i8v8double
|
||||
; SSE2: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv8i8v8double
|
||||
; AVX1-LABEL: sitofpv8i8v8double
|
||||
; AVX1: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv8i8v8double
|
||||
; AVX2-LABEL: sitofpv8i8v8double
|
||||
; AVX2: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv8i8v8double
|
||||
; AVX512F-LABEL: sitofpv8i8v8double
|
||||
; AVX512F: cost of 2 {{.*}} sitofp
|
||||
%1 = sitofp <8 x i8> %a to <8 x double>
|
||||
ret <8 x double> %1
|
||||
}
|
||||
|
||||
define <16 x double> @sitofpv16i8v16double(<16 x i8> %a) {
|
||||
; SSE2: sitofpv16i8v16double
|
||||
; SSE2-LABEL: sitofpv16i8v16double
|
||||
; SSE2: cost of 160 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv16i8v16double
|
||||
; AVX1-LABEL: sitofpv16i8v16double
|
||||
; AVX1: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv16i8v16double
|
||||
; AVX2-LABEL: sitofpv16i8v16double
|
||||
; AVX2: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv16i8v16double
|
||||
; AVX512F-LABEL: sitofpv16i8v16double
|
||||
; AVX512F: cost of 44 {{.*}} sitofp
|
||||
%1 = sitofp <16 x i8> %a to <16 x double>
|
||||
ret <16 x double> %1
|
||||
}
|
||||
|
||||
define <32 x double> @sitofpv32i8v32double(<32 x i8> %a) {
|
||||
; SSE2: sitofpv32i8v32double
|
||||
; SSE2-LABEL: sitofpv32i8v32double
|
||||
; SSE2: cost of 320 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv32i8v32double
|
||||
; AVX1-LABEL: sitofpv32i8v32double
|
||||
; AVX1: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv32i8v32double
|
||||
; AVX2-LABEL: sitofpv32i8v32double
|
||||
; AVX2: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv32i8v32double
|
||||
; AVX512F-LABEL: sitofpv32i8v32double
|
||||
; AVX512F: cost of 88 {{.*}} sitofp
|
||||
%1 = sitofp <32 x i8> %a to <32 x double>
|
||||
ret <32 x double> %1
|
||||
}
|
||||
|
||||
define <2 x double> @sitofpv2i16v2double(<2 x i16> %a) {
|
||||
; SSE2: sitofpv2i16v2double
|
||||
; SSE2-LABEL: sitofpv2i16v2double
|
||||
; SSE2: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv2i16v2double
|
||||
; AVX1-LABEL: sitofpv2i16v2double
|
||||
; AVX1: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv2i16v2double
|
||||
; AVX2-LABEL: sitofpv2i16v2double
|
||||
; AVX2: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv2i16v2double
|
||||
; AVX512F-LABEL: sitofpv2i16v2double
|
||||
; AVX512F: cost of 4 {{.*}} sitofp
|
||||
%1 = sitofp <2 x i16> %a to <2 x double>
|
||||
ret <2 x double> %1
|
||||
}
|
||||
|
||||
define <4 x double> @sitofpv4i16v4double(<4 x i16> %a) {
|
||||
; SSE2: sitofpv4i16v4double
|
||||
; SSE2-LABEL: sitofpv4i16v4double
|
||||
; SSE2: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv4i16v4double
|
||||
; AVX1-LABEL: sitofpv4i16v4double
|
||||
; AVX1: cost of 3 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv4i16v4double
|
||||
; AVX2-LABEL: sitofpv4i16v4double
|
||||
; AVX2: cost of 3 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv4i16v4double
|
||||
; AVX512F-LABEL: sitofpv4i16v4double
|
||||
; AVX512F: cost of 3 {{.*}} sitofp
|
||||
%1 = sitofp <4 x i16> %a to <4 x double>
|
||||
ret <4 x double> %1
|
||||
}
|
||||
|
||||
define <8 x double> @sitofpv8i16v8double(<8 x i16> %a) {
|
||||
; SSE2: sitofpv8i16v8double
|
||||
; SSE2-LABEL: sitofpv8i16v8double
|
||||
; SSE2: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv8i16v8double
|
||||
; AVX1-LABEL: sitofpv8i16v8double
|
||||
; AVX1: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv8i16v8double
|
||||
; AVX2-LABEL: sitofpv8i16v8double
|
||||
; AVX2: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv8i16v8double
|
||||
; AVX512F-LABEL: sitofpv8i16v8double
|
||||
; AVX512F: cost of 2 {{.*}} sitofp
|
||||
%1 = sitofp <8 x i16> %a to <8 x double>
|
||||
ret <8 x double> %1
|
||||
}
|
||||
|
||||
define <16 x double> @sitofpv16i16v16double(<16 x i16> %a) {
|
||||
; SSE2: sitofpv16i16v16double
|
||||
; SSE2-LABEL: sitofpv16i16v16double
|
||||
; SSE2: cost of 160 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv16i16v16double
|
||||
; AVX1-LABEL: sitofpv16i16v16double
|
||||
; AVX1: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv16i16v16double
|
||||
; AVX2-LABEL: sitofpv16i16v16double
|
||||
; AVX2: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv16i16v16double
|
||||
; AVX512F-LABEL: sitofpv16i16v16double
|
||||
; AVX512F: cost of 44 {{.*}} sitofp
|
||||
%1 = sitofp <16 x i16> %a to <16 x double>
|
||||
ret <16 x double> %1
|
||||
}
|
||||
|
||||
define <32 x double> @sitofpv32i16v32double(<32 x i16> %a) {
|
||||
; SSE2: sitofpv32i16v32double
|
||||
; SSE2-LABEL: sitofpv32i16v32double
|
||||
; SSE2: cost of 320 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv32i16v32double
|
||||
; AVX1-LABEL: sitofpv32i16v32double
|
||||
; AVX1: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv32i16v32double
|
||||
; AVX2-LABEL: sitofpv32i16v32double
|
||||
; AVX2: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv32i16v32double
|
||||
; AVX512F-LABEL: sitofpv32i16v32double
|
||||
; AVX512F: cost of 88 {{.*}} sitofp
|
||||
%1 = sitofp <32 x i16> %a to <32 x double>
|
||||
ret <32 x double> %1
|
||||
}
|
||||
|
||||
define <2 x double> @sitofpv2i32v2double(<2 x i32> %a) {
|
||||
; SSE2: sitofpv2i32v2double
|
||||
; SSE2-LABEL: sitofpv2i32v2double
|
||||
; SSE2: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv2i32v2double
|
||||
; AVX1-LABEL: sitofpv2i32v2double
|
||||
; AVX1: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv2i32v2double
|
||||
; AVX2-LABEL: sitofpv2i32v2double
|
||||
; AVX2: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv2i32v2double
|
||||
; AVX512F-LABEL: sitofpv2i32v2double
|
||||
; AVX512F: cost of 4 {{.*}} sitofp
|
||||
%1 = sitofp <2 x i32> %a to <2 x double>
|
||||
ret <2 x double> %1
|
||||
}
|
||||
|
||||
define <4 x double> @sitofpv4i32v4double(<4 x i32> %a) {
|
||||
; SSE2: sitofpv4i32v4double
|
||||
; SSE2-LABEL: sitofpv4i32v4double
|
||||
; SSE2: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv4i32v4double
|
||||
; AVX1-LABEL: sitofpv4i32v4double
|
||||
; AVX1: cost of 1 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv4i32v4double
|
||||
; AVX2-LABEL: sitofpv4i32v4double
|
||||
; AVX2: cost of 1 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv4i32v4double
|
||||
; AVX512F-LABEL: sitofpv4i32v4double
|
||||
; AVX512F: cost of 1 {{.*}} sitofp
|
||||
%1 = sitofp <4 x i32> %a to <4 x double>
|
||||
ret <4 x double> %1
|
||||
}
|
||||
|
||||
define <8 x double> @sitofpv8i32v8double(<8 x i32> %a) {
|
||||
; SSE2: sitofpv8i32v8double
|
||||
; SSE2-LABEL: sitofpv8i32v8double
|
||||
; SSE2: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv8i32v8double
|
||||
; AVX1-LABEL: sitofpv8i32v8double
|
||||
; AVX1: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv8i32v8double
|
||||
; AVX2-LABEL: sitofpv8i32v8double
|
||||
; AVX2: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv8i32v8double
|
||||
; AVX512F-LABEL: sitofpv8i32v8double
|
||||
; AVX512F: cost of 1 {{.*}} sitofp
|
||||
%1 = sitofp <8 x i32> %a to <8 x double>
|
||||
ret <8 x double> %1
|
||||
}
|
||||
|
||||
define <16 x double> @sitofpv16i32v16double(<16 x i32> %a) {
|
||||
; SSE2: sitofpv16i32v16double
|
||||
; SSE2-LABEL: sitofpv16i32v16double
|
||||
; SSE2: cost of 160 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv16i32v16double
|
||||
; AVX1-LABEL: sitofpv16i32v16double
|
||||
; AVX1: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv16i32v16double
|
||||
; AVX2-LABEL: sitofpv16i32v16double
|
||||
; AVX2: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv16i32v16double
|
||||
; AVX512F-LABEL: sitofpv16i32v16double
|
||||
; AVX512F: cost of 44 {{.*}} sitofp
|
||||
%1 = sitofp <16 x i32> %a to <16 x double>
|
||||
ret <16 x double> %1
|
||||
}
|
||||
|
||||
define <32 x double> @sitofpv32i32v32double(<32 x i32> %a) {
|
||||
; SSE2: sitofpv32i32v32double
|
||||
; SSE2-LABEL: sitofpv32i32v32double
|
||||
; SSE2: cost of 320 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv32i32v32double
|
||||
; AVX1-LABEL: sitofpv32i32v32double
|
||||
; AVX1: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv32i32v32double
|
||||
; AVX2-LABEL: sitofpv32i32v32double
|
||||
; AVX2: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv32i32v32double
|
||||
; AVX512F-LABEL: sitofpv32i32v32double
|
||||
; AVX512F: cost of 88 {{.*}} sitofp
|
||||
%1 = sitofp <32 x i32> %a to <32 x double>
|
||||
ret <32 x double> %1
|
||||
}
|
||||
|
||||
define <2 x double> @sitofpv2i64v2double(<2 x i64> %a) {
|
||||
; SSE2: sitofpv2i64v2double
|
||||
; SSE2-LABEL: sitofpv2i64v2double
|
||||
; SSE2: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv2i64v2double
|
||||
; AVX1-LABEL: sitofpv2i64v2double
|
||||
; AVX1: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv2i64v2double
|
||||
; AVX2-LABEL: sitofpv2i64v2double
|
||||
; AVX2: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv2i64v2double
|
||||
; AVX512F-LABEL: sitofpv2i64v2double
|
||||
; AVX512F: cost of 4 {{.*}} sitofp
|
||||
%1 = sitofp <2 x i64> %a to <2 x double>
|
||||
ret <2 x double> %1
|
||||
}
|
||||
|
||||
define <4 x double> @sitofpv4i64v4double(<4 x i64> %a) {
|
||||
; SSE2: sitofpv4i64v4double
|
||||
; SSE2-LABEL: sitofpv4i64v4double
|
||||
; SSE2: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv4i64v4double
|
||||
; AVX1-LABEL: sitofpv4i64v4double
|
||||
; AVX1: cost of 10 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv4i64v4double
|
||||
; AVX2-LABEL: sitofpv4i64v4double
|
||||
; AVX2: cost of 10 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv4i64v4double
|
||||
; AVX512F-LABEL: sitofpv4i64v4double
|
||||
; AVX512F: cost of 10 {{.*}} sitofp
|
||||
%1 = sitofp <4 x i64> %a to <4 x double>
|
||||
ret <4 x double> %1
|
||||
}
|
||||
|
||||
define <8 x double> @sitofpv8i64v8double(<8 x i64> %a) {
|
||||
; SSE2: sitofpv8i64v8double
|
||||
; SSE2-LABEL: sitofpv8i64v8double
|
||||
; SSE2: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv8i64v8double
|
||||
; AVX1-LABEL: sitofpv8i64v8double
|
||||
; AVX1: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv8i64v8double
|
||||
; AVX2-LABEL: sitofpv8i64v8double
|
||||
; AVX2: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv8i64v8double
|
||||
; AVX512F-LABEL: sitofpv8i64v8double
|
||||
; AVX512F: cost of 22 {{.*}} sitofp
|
||||
%1 = sitofp <8 x i64> %a to <8 x double>
|
||||
ret <8 x double> %1
|
||||
}
|
||||
|
||||
define <16 x double> @sitofpv16i64v16double(<16 x i64> %a) {
|
||||
; SSE2: sitofpv16i64v16double
|
||||
; SSE2-LABEL: sitofpv16i64v16double
|
||||
; SSE2: cost of 160 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv16i64v16double
|
||||
; AVX1-LABEL: sitofpv16i64v16double
|
||||
; AVX1: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv16i64v16double
|
||||
; AVX2-LABEL: sitofpv16i64v16double
|
||||
; AVX2: cost of 40 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv16i64v16double
|
||||
; AVX512F-LABEL: sitofpv16i64v16double
|
||||
; AVX512F: cost of 44 {{.*}} sitofp
|
||||
%1 = sitofp <16 x i64> %a to <16 x double>
|
||||
ret <16 x double> %1
|
||||
}
|
||||
|
||||
define <32 x double> @sitofpv32i64v32double(<32 x i64> %a) {
|
||||
; SSE2: sitofpv32i64v32double
|
||||
; SSE2-LABEL: sitofpv32i64v32double
|
||||
; SSE2: cost of 320 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv32i64v32double
|
||||
; AVX1-LABEL: sitofpv32i64v32double
|
||||
; AVX1: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv32i64v32double
|
||||
; AVX2-LABEL: sitofpv32i64v32double
|
||||
; AVX2: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv32i64v32double
|
||||
; AVX512F-LABEL: sitofpv32i64v32double
|
||||
; AVX512F: cost of 88 {{.*}} sitofp
|
||||
%1 = sitofp <32 x i64> %a to <32 x double>
|
||||
ret <32 x double> %1
|
||||
}
|
||||
|
||||
define <2 x float> @sitofpv2i8v2float(<2 x i8> %a) {
|
||||
; SSE2: sitofpv2i8v2float
|
||||
; SSE2-LABEL: sitofpv2i8v2float
|
||||
; SSE2: cost of 15 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv2i8v2float
|
||||
; AVX1-LABEL: sitofpv2i8v2float
|
||||
; AVX1: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv2i8v2float
|
||||
; AVX2-LABEL: sitofpv2i8v2float
|
||||
; AVX2: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv2i8v2float
|
||||
; AVX512F-LABEL: sitofpv2i8v2float
|
||||
; AVX512F: cost of 4 {{.*}} sitofp
|
||||
%1 = sitofp <2 x i8> %a to <2 x float>
|
||||
ret <2 x float> %1
|
||||
}
|
||||
|
||||
define <4 x float> @sitofpv4i8v4float(<4 x i8> %a) {
|
||||
; SSE2: sitofpv4i8v4float
|
||||
; SSE2-LABEL: sitofpv4i8v4float
|
||||
; SSE2: cost of 15 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv4i8v4float
|
||||
; AVX1-LABEL: sitofpv4i8v4float
|
||||
; AVX1: cost of 3 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv4i8v4float
|
||||
; AVX2-LABEL: sitofpv4i8v4float
|
||||
; AVX2: cost of 3 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv4i8v4float
|
||||
; AVX512F-LABEL: sitofpv4i8v4float
|
||||
; AVX512F: cost of 3 {{.*}} sitofp
|
||||
%1 = sitofp <4 x i8> %a to <4 x float>
|
||||
ret <4 x float> %1
|
||||
}
|
||||
|
||||
define <8 x float> @sitofpv8i8v8float(<8 x i8> %a) {
|
||||
; SSE2: sitofpv8i8v8float
|
||||
; SSE2-LABEL: sitofpv8i8v8float
|
||||
; SSE2: cost of 15 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv8i8v8float
|
||||
; AVX1-LABEL: sitofpv8i8v8float
|
||||
; AVX1: cost of 8 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv8i8v8float
|
||||
; AVX2-LABEL: sitofpv8i8v8float
|
||||
; AVX2: cost of 8 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv8i8v8float
|
||||
; AVX512F-LABEL: sitofpv8i8v8float
|
||||
; AVX512F: cost of 8 {{.*}} sitofp
|
||||
%1 = sitofp <8 x i8> %a to <8 x float>
|
||||
ret <8 x float> %1
|
||||
}
|
||||
|
||||
define <16 x float> @sitofpv16i8v16float(<16 x i8> %a) {
|
||||
; SSE2: sitofpv16i8v16float
|
||||
; SSE2-LABEL: sitofpv16i8v16float
|
||||
; SSE2: cost of 8 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv16i8v16float
|
||||
; AVX1-LABEL: sitofpv16i8v16float
|
||||
; AVX1: cost of 44 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv16i8v16float
|
||||
; AVX2-LABEL: sitofpv16i8v16float
|
||||
; AVX2: cost of 44 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv16i8v16float
|
||||
; AVX512F-LABEL: sitofpv16i8v16float
|
||||
; AVX512F: cost of 2 {{.*}} sitofp
|
||||
%1 = sitofp <16 x i8> %a to <16 x float>
|
||||
ret <16 x float> %1
|
||||
}
|
||||
|
||||
define <32 x float> @sitofpv32i8v32float(<32 x i8> %a) {
|
||||
; SSE2: sitofpv32i8v32float
|
||||
; SSE2-LABEL: sitofpv32i8v32float
|
||||
; SSE2: cost of 16 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv32i8v32float
|
||||
; AVX1-LABEL: sitofpv32i8v32float
|
||||
; AVX1: cost of 88 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv32i8v32float
|
||||
; AVX2-LABEL: sitofpv32i8v32float
|
||||
; AVX2: cost of 88 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv32i8v32float
|
||||
; AVX512F-LABEL: sitofpv32i8v32float
|
||||
; AVX512F: cost of 92 {{.*}} sitofp
|
||||
%1 = sitofp <32 x i8> %a to <32 x float>
|
||||
ret <32 x float> %1
|
||||
}
|
||||
|
||||
define <2 x float> @sitofpv2i16v2float(<2 x i16> %a) {
|
||||
; SSE2: sitofpv2i16v2float
|
||||
; SSE2-LABEL: sitofpv2i16v2float
|
||||
; SSE2: cost of 15 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv2i16v2float
|
||||
; AVX1-LABEL: sitofpv2i16v2float
|
||||
; AVX1: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv2i16v2float
|
||||
; AVX2-LABEL: sitofpv2i16v2float
|
||||
; AVX2: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv2i16v2float
|
||||
; AVX512F-LABEL: sitofpv2i16v2float
|
||||
; AVX512F: cost of 4 {{.*}} sitofp
|
||||
%1 = sitofp <2 x i16> %a to <2 x float>
|
||||
ret <2 x float> %1
|
||||
}
|
||||
|
||||
define <4 x float> @sitofpv4i16v4float(<4 x i16> %a) {
|
||||
; SSE2: sitofpv4i16v4float
|
||||
; SSE2-LABEL: sitofpv4i16v4float
|
||||
; SSE2: cost of 15 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv4i16v4float
|
||||
; AVX1-LABEL: sitofpv4i16v4float
|
||||
; AVX1: cost of 3 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv4i16v4float
|
||||
; AVX2-LABEL: sitofpv4i16v4float
|
||||
; AVX2: cost of 3 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv4i16v4float
|
||||
; AVX512F-LABEL: sitofpv4i16v4float
|
||||
; AVX512F: cost of 3 {{.*}} sitofp
|
||||
%1 = sitofp <4 x i16> %a to <4 x float>
|
||||
ret <4 x float> %1
|
||||
}
|
||||
|
||||
define <8 x float> @sitofpv8i16v8float(<8 x i16> %a) {
|
||||
; SSE2: sitofpv8i16v8float
|
||||
; SSE2-LABEL: sitofpv8i16v8float
|
||||
; SSE2: cost of 15 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv8i16v8float
|
||||
; AVX1-LABEL: sitofpv8i16v8float
|
||||
; AVX1: cost of 5 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv8i16v8float
|
||||
; AVX2-LABEL: sitofpv8i16v8float
|
||||
; AVX2: cost of 5 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv8i16v8float
|
||||
; AVX512F-LABEL: sitofpv8i16v8float
|
||||
; AVX512F: cost of 5 {{.*}} sitofp
|
||||
%1 = sitofp <8 x i16> %a to <8 x float>
|
||||
ret <8 x float> %1
|
||||
}
|
||||
|
||||
define <16 x float> @sitofpv16i16v16float(<16 x i16> %a) {
|
||||
; SSE2: sitofpv16i16v16float
|
||||
; SSE2-LABEL: sitofpv16i16v16float
|
||||
; SSE2: cost of 30 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv16i16v16float
|
||||
; AVX1-LABEL: sitofpv16i16v16float
|
||||
; AVX1: cost of 44 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv16i16v16float
|
||||
; AVX2-LABEL: sitofpv16i16v16float
|
||||
; AVX2: cost of 44 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv16i16v16float
|
||||
; AVX512F-LABEL: sitofpv16i16v16float
|
||||
; AVX512F: cost of 2 {{.*}} sitofp
|
||||
%1 = sitofp <16 x i16> %a to <16 x float>
|
||||
ret <16 x float> %1
|
||||
}
|
||||
|
||||
define <32 x float> @sitofpv32i16v32float(<32 x i16> %a) {
|
||||
; SSE2: sitofpv32i16v32float
|
||||
; SSE2-LABEL: sitofpv32i16v32float
|
||||
; SSE2: cost of 60 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv32i16v32float
|
||||
; AVX1-LABEL: sitofpv32i16v32float
|
||||
; AVX1: cost of 88 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv32i16v32float
|
||||
; AVX2-LABEL: sitofpv32i16v32float
|
||||
; AVX2: cost of 88 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv32i16v32float
|
||||
; AVX512F: cost of 2 {{.*}} sitofp
|
||||
; AVX512F-LABEL: sitofpv32i16v32float
|
||||
; AVX512F: cost of 92 {{.*}} sitofp
|
||||
%1 = sitofp <32 x i16> %a to <32 x float>
|
||||
ret <32 x float> %1
|
||||
}
|
||||
|
||||
define <2 x float> @sitofpv2i32v2float(<2 x i32> %a) {
|
||||
; SSE2: sitofpv2i32v2float
|
||||
; SSE2-LABEL: sitofpv2i32v2float
|
||||
; SSE2: cost of 15 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv2i32v2float
|
||||
; AVX1-LABEL: sitofpv2i32v2float
|
||||
; AVX1: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv2i32v2float
|
||||
; AVX2-LABEL: sitofpv2i32v2float
|
||||
; AVX2: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv2i32v2float
|
||||
; AVX512F-LABEL: sitofpv2i32v2float
|
||||
; AVX512F: cost of 4 {{.*}} sitofp
|
||||
%1 = sitofp <2 x i32> %a to <2 x float>
|
||||
ret <2 x float> %1
|
||||
}
|
||||
|
||||
define <4 x float> @sitofpv4i32v4float(<4 x i32> %a) {
|
||||
; SSE2: sitofpv4i32v4float
|
||||
; SSE2-LABEL: sitofpv4i32v4float
|
||||
; SSE2: cost of 15 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv4i32v4float
|
||||
; AVX1-LABEL: sitofpv4i32v4float
|
||||
; AVX1: cost of 1 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv4i32v4float
|
||||
; AVX2-LABEL: sitofpv4i32v4float
|
||||
; AVX2: cost of 1 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv4i32v4float
|
||||
; AVX512F-LABEL: sitofpv4i32v4float
|
||||
; AVX512F: cost of 1 {{.*}} sitofp
|
||||
%1 = sitofp <4 x i32> %a to <4 x float>
|
||||
ret <4 x float> %1
|
||||
}
|
||||
|
||||
define <8 x float> @sitofpv8i32v8float(<8 x i32> %a) {
|
||||
; SSE2: sitofpv8i32v8float
|
||||
; SSE2-LABEL: sitofpv8i32v8float
|
||||
; SSE2: cost of 30 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv8i32v8float
|
||||
; AVX1-LABEL: sitofpv8i32v8float
|
||||
; AVX1: cost of 1 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv8i32v8float
|
||||
; AVX2-LABEL: sitofpv8i32v8float
|
||||
; AVX2: cost of 1 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv8i32v8float
|
||||
; AVX512F-LABEL: sitofpv8i32v8float
|
||||
; AVX512F: cost of 1 {{.*}} sitofp
|
||||
%1 = sitofp <8 x i32> %a to <8 x float>
|
||||
ret <8 x float> %1
|
||||
}
|
||||
|
||||
define <16 x float> @sitofpv16i32v16float(<16 x i32> %a) {
|
||||
; SSE2: sitofpv16i32v16float
|
||||
; SSE2-LABEL: sitofpv16i32v16float
|
||||
; SSE2: cost of 60 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv16i32v16float
|
||||
; AVX1-LABEL: sitofpv16i32v16float
|
||||
; AVX1: cost of 44 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv16i32v16float
|
||||
; AVX2-LABEL: sitofpv16i32v16float
|
||||
; AVX2: cost of 44 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv16i32v16float
|
||||
; AVX512F-LABEL: sitofpv16i32v16float
|
||||
; AVX512F: cost of 1 {{.*}} sitofp
|
||||
%1 = sitofp <16 x i32> %a to <16 x float>
|
||||
ret <16 x float> %1
|
||||
}
|
||||
|
||||
define <32 x float> @sitofpv32i32v32float(<32 x i32> %a) {
|
||||
; SSE2: sitofpv32i32v32float
|
||||
; SSE2-LABEL: sitofpv32i32v32float
|
||||
; SSE2: cost of 120 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv32i32v32float
|
||||
; AVX1-LABEL: sitofpv32i32v32float
|
||||
; AVX1: cost of 88 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv32i32v32float
|
||||
; AVX2-LABEL: sitofpv32i32v32float
|
||||
; AVX2: cost of 88 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv32i32v32float
|
||||
; AVX512F: cost of 1 {{.*}} sitofp
|
||||
; AVX512F-LABEL: sitofpv32i32v32float
|
||||
; AVX512F: cost of 92 {{.*}} sitofp
|
||||
%1 = sitofp <32 x i32> %a to <32 x float>
|
||||
ret <32 x float> %1
|
||||
}
|
||||
|
||||
define <2 x float> @sitofpv2i64v2float(<2 x i64> %a) {
|
||||
; SSE2: sitofpv2i64v2float
|
||||
; SSE2-LABEL: sitofpv2i64v2float
|
||||
; SSE2: cost of 15 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv2i64v2float
|
||||
; AVX1-LABEL: sitofpv2i64v2float
|
||||
; AVX1: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv2i64v2float
|
||||
; AVX2-LABEL: sitofpv2i64v2float
|
||||
; AVX2: cost of 4 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv2i64v2float
|
||||
; AVX512F-LABEL: sitofpv2i64v2float
|
||||
; AVX512F: cost of 4 {{.*}} sitofp
|
||||
%1 = sitofp <2 x i64> %a to <2 x float>
|
||||
ret <2 x float> %1
|
||||
}
|
||||
|
||||
define <4 x float> @sitofpv4i64v4float(<4 x i64> %a) {
|
||||
; SSE2: sitofpv4i64v4float
|
||||
; SSE2-LABEL: sitofpv4i64v4float
|
||||
; SSE2: cost of 30 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv4i64v4float
|
||||
; AVX1-LABEL: sitofpv4i64v4float
|
||||
; AVX1: cost of 10 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv4i64v4float
|
||||
; AVX2-LABEL: sitofpv4i64v4float
|
||||
; AVX2: cost of 10 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv4i64v4float
|
||||
; AVX512F-LABEL: sitofpv4i64v4float
|
||||
; AVX512F: cost of 10 {{.*}} sitofp
|
||||
%1 = sitofp <4 x i64> %a to <4 x float>
|
||||
ret <4 x float> %1
|
||||
}
|
||||
|
||||
define <8 x float> @sitofpv8i64v8float(<8 x i64> %a) {
|
||||
; SSE2: sitofpv8i64v8float
|
||||
; SSE2-LABEL: sitofpv8i64v8float
|
||||
; SSE2: cost of 60 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv8i64v8float
|
||||
; AVX1-LABEL: sitofpv8i64v8float
|
||||
; AVX1: cost of 22 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv8i64v8float
|
||||
; AVX2-LABEL: sitofpv8i64v8float
|
||||
; AVX2: cost of 22 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv8i64v8float
|
||||
; AVX512F-LABEL: sitofpv8i64v8float
|
||||
; AVX512F: cost of 22 {{.*}} sitofp
|
||||
%1 = sitofp <8 x i64> %a to <8 x float>
|
||||
ret <8 x float> %1
|
||||
}
|
||||
|
||||
define <16 x float> @sitofpv16i64v16float(<16 x i64> %a) {
|
||||
; SSE2: sitofpv16i64v16float
|
||||
; SSE2-LABEL: sitofpv16i64v16float
|
||||
; SSE2: cost of 120 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv16i64v16float
|
||||
; AVX1-LABEL: sitofpv16i64v16float
|
||||
; AVX1: cost of 44 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv16i64v16float
|
||||
; AVX2-LABEL: sitofpv16i64v16float
|
||||
; AVX2: cost of 44 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv16i64v16float
|
||||
; AVX512F-LABEL: sitofpv16i64v16float
|
||||
; AVX512F: cost of 46 {{.*}} sitofp
|
||||
%1 = sitofp <16 x i64> %a to <16 x float>
|
||||
ret <16 x float> %1
|
||||
}
|
||||
|
||||
define <32 x float> @sitofpv32i64v32float(<32 x i64> %a) {
|
||||
; SSE2: sitofpv32i64v32float
|
||||
; SSE2-LABEL: sitofpv32i64v32float
|
||||
; SSE2: cost of 240 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv32i64v32float
|
||||
; AVX1-LABEL: sitofpv32i64v32float
|
||||
; AVX1: cost of 88 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv32i64v32float
|
||||
; AVX2-LABEL: sitofpv32i64v32float
|
||||
; AVX2: cost of 88 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv32i64v32float
|
||||
; AVX512F-LABEL: sitofpv32i64v32float
|
||||
; AVX512F: cost of 92 {{.*}} sitofp
|
||||
%1 = sitofp <32 x i64> %a to <32 x float>
|
||||
ret <32 x float> %1
|
||||
}
|
||||
|
||||
define <8 x double> @sitofpv8i1v8double(<8 x double> %a) {
|
||||
; SSE2: sitofpv8i1v8double
|
||||
; SSE2-LABEL: sitofpv8i1v8double
|
||||
; SSE2: cost of 80 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv8i1v8double
|
||||
; AVX1-LABEL: sitofpv8i1v8double
|
||||
; AVX1: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv8i1v8double
|
||||
; AVX2-LABEL: sitofpv8i1v8double
|
||||
; AVX2: cost of 20 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv8i1v8double
|
||||
; AVX512F-LABEL: sitofpv8i1v8double
|
||||
; AVX512F: cost of 4 {{.*}} sitofp
|
||||
%cmpres = fcmp ogt <8 x double> %a, zeroinitializer
|
||||
%1 = sitofp <8 x i1> %cmpres to <8 x double>
|
||||
@ -661,16 +661,16 @@ define <8 x double> @sitofpv8i1v8double(<8 x double> %a) {
|
||||
}
|
||||
|
||||
define <16 x float> @sitofpv16i1v16float(<16 x float> %a) {
|
||||
; SSE2: sitofpv16i1v16float
|
||||
; SSE2-LABEL: sitofpv16i1v16float
|
||||
; SSE2: cost of 8 {{.*}} sitofp
|
||||
;
|
||||
; AVX1: sitofpv16i1v16float
|
||||
; AVX1-LABEL: sitofpv16i1v16float
|
||||
; AVX1: cost of 44 {{.*}} sitofp
|
||||
;
|
||||
; AVX2: sitofpv16i1v16float
|
||||
; AVX2-LABEL: sitofpv16i1v16float
|
||||
; AVX2: cost of 44 {{.*}} sitofp
|
||||
;
|
||||
; AVX512F: sitofpv16i1v16float
|
||||
; AVX512F-LABEL: sitofpv16i1v16float
|
||||
; AVX512F: cost of 3 {{.*}} sitofp
|
||||
%cmpres = fcmp ogt <16 x float> %a, zeroinitializer
|
||||
%1 = sitofp <16 x i1> %cmpres to <16 x float>
|
||||
|
||||
@ -2,644 +2,708 @@
|
||||
; RUN: opt -mtriple=x86_64-apple-darwin -mattr=+avx -cost-model -analyze < %s | FileCheck --check-prefix=AVX --check-prefix=AVX1 %s
|
||||
; RUN: opt -mtriple=x86_64-apple-darwin -mattr=+avx2 -cost-model -analyze < %s | FileCheck --check-prefix=AVX --check-prefix=AVX2 %s
|
||||
; RUN: opt -mtriple=x86_64-apple-darwin -mattr=+avx512f -cost-model -analyze < %s | FileCheck --check-prefix=AVX512F %s
|
||||
; RUN: opt -mtriple=x86_64-apple-darwin -mattr=+avx512dq -cost-model -analyze < %s | FileCheck --check-prefix=AVX512DQ %s
|
||||
|
||||
define <2 x double> @uitofpv2i8v2double(<2 x i8> %a) {
|
||||
; SSE2: uitofpv2i8v2double
|
||||
; SSE2-LABEL: uitofpv2i8v2double
|
||||
; SSE2: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv2i8v2double
|
||||
; AVX1-LABEL: uitofpv2i8v2double
|
||||
; AVX1: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv2i8v2double
|
||||
; AVX2-LABEL: uitofpv2i8v2double
|
||||
; AVX2: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv2i8v2double
|
||||
; AVX512F: cost of 4 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv2i8v2double
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <2 x i8> %a to <2 x double>
|
||||
ret <2 x double> %1
|
||||
}
|
||||
|
||||
define <4 x double> @uitofpv4i8v4double(<4 x i8> %a) {
|
||||
; SSE2: uitofpv4i8v4double
|
||||
; SSE2-LABEL: uitofpv4i8v4double
|
||||
; SSE2: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv4i8v4double
|
||||
; AVX1-LABEL: uitofpv4i8v4double
|
||||
; AVX1: cost of 2 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv4i8v4double
|
||||
; AVX2-LABEL: uitofpv4i8v4double
|
||||
; AVX2: cost of 2 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv4i8v4double
|
||||
; AVX512F-LABEL: uitofpv4i8v4double
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <4 x i8> %a to <4 x double>
|
||||
ret <4 x double> %1
|
||||
}
|
||||
|
||||
define <8 x double> @uitofpv8i8v8double(<8 x i8> %a) {
|
||||
; SSE2: uitofpv8i8v8double
|
||||
; SSE2-LABEL: uitofpv8i8v8double
|
||||
; SSE2: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv8i8v8double
|
||||
; AVX1-LABEL: uitofpv8i8v8double
|
||||
; AVX1: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv8i8v8double
|
||||
; AVX2-LABEL: uitofpv8i8v8double
|
||||
; AVX2: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv8i8v8double
|
||||
; AVX512F: cost of 22 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv8i8v8double
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <8 x i8> %a to <8 x double>
|
||||
ret <8 x double> %1
|
||||
}
|
||||
|
||||
define <16 x double> @uitofpv16i8v16double(<16 x i8> %a) {
|
||||
; SSE2: uitofpv16i8v16double
|
||||
; SSE2-LABEL: uitofpv16i8v16double
|
||||
; SSE2: cost of 160 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv16i8v16double
|
||||
; AVX1-LABEL: uitofpv16i8v16double
|
||||
; AVX1: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv16i8v16double
|
||||
; AVX2-LABEL: uitofpv16i8v16double
|
||||
; AVX2: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv16i8v16double
|
||||
; AVX512F-LABEL: uitofpv16i8v16double
|
||||
; AVX512F: cost of 44 {{.*}} uitofp
|
||||
%1 = uitofp <16 x i8> %a to <16 x double>
|
||||
ret <16 x double> %1
|
||||
}
|
||||
|
||||
define <32 x double> @uitofpv32i8v32double(<32 x i8> %a) {
|
||||
; SSE2: uitofpv32i8v32double
|
||||
; SSE2-LABEL: uitofpv32i8v32double
|
||||
; SSE2: cost of 320 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv32i8v32double
|
||||
; AVX1-LABEL: uitofpv32i8v32double
|
||||
; AVX1: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv32i8v32double
|
||||
; AVX2-LABEL: uitofpv32i8v32double
|
||||
; AVX2: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv32i8v32double
|
||||
; AVX512F-LABEL: uitofpv32i8v32double
|
||||
; AVX512F: cost of 88 {{.*}} uitofp
|
||||
%1 = uitofp <32 x i8> %a to <32 x double>
|
||||
ret <32 x double> %1
|
||||
}
|
||||
|
||||
define <2 x double> @uitofpv2i16v2double(<2 x i16> %a) {
|
||||
; SSE2: uitofpv2i16v2double
|
||||
; SSE2-LABEL: uitofpv2i16v2double
|
||||
; SSE2: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv2i16v2double
|
||||
; AVX1-LABEL: uitofpv2i16v2double
|
||||
; AVX1: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv2i16v2double
|
||||
; AVX2-LABEL: uitofpv2i16v2double
|
||||
; AVX2: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv2i16v2double
|
||||
; AVX512F: cost of 4 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv2i16v2double
|
||||
; AVX512F: cost of 5 {{.*}} uitofp
|
||||
%1 = uitofp <2 x i16> %a to <2 x double>
|
||||
ret <2 x double> %1
|
||||
}
|
||||
|
||||
define <4 x double> @uitofpv4i16v4double(<4 x i16> %a) {
|
||||
; SSE2: uitofpv4i16v4double
|
||||
; SSE2-LABEL: uitofpv4i16v4double
|
||||
; SSE2: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv4i16v4double
|
||||
; AVX1-LABEL: uitofpv4i16v4double
|
||||
; AVX1: cost of 2 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv4i16v4double
|
||||
; AVX2-LABEL: uitofpv4i16v4double
|
||||
; AVX2: cost of 2 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv4i16v4double
|
||||
; AVX512F-LABEL: uitofpv4i16v4double
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <4 x i16> %a to <4 x double>
|
||||
ret <4 x double> %1
|
||||
}
|
||||
|
||||
define <8 x double> @uitofpv8i16v8double(<8 x i16> %a) {
|
||||
; SSE2: uitofpv8i16v8double
|
||||
; SSE2-LABEL: uitofpv8i16v8double
|
||||
; SSE2: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv8i16v8double
|
||||
; AVX1-LABEL: uitofpv8i16v8double
|
||||
; AVX1: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv8i16v8double
|
||||
; AVX2-LABEL: uitofpv8i16v8double
|
||||
; AVX2: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv8i16v8double
|
||||
; AVX512F: cost of 22 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv8i16v8double
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <8 x i16> %a to <8 x double>
|
||||
ret <8 x double> %1
|
||||
}
|
||||
|
||||
define <16 x double> @uitofpv16i16v16double(<16 x i16> %a) {
|
||||
; SSE2: uitofpv16i16v16double
|
||||
; SSE2-LABEL: uitofpv16i16v16double
|
||||
; SSE2: cost of 160 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv16i16v16double
|
||||
; AVX1-LABEL: uitofpv16i16v16double
|
||||
; AVX1: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv16i16v16double
|
||||
; AVX2-LABEL: uitofpv16i16v16double
|
||||
; AVX2: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv16i16v16double
|
||||
; AVX512F-LABEL: uitofpv16i16v16double
|
||||
; AVX512F: cost of 44 {{.*}} uitofp
|
||||
%1 = uitofp <16 x i16> %a to <16 x double>
|
||||
ret <16 x double> %1
|
||||
}
|
||||
|
||||
define <32 x double> @uitofpv32i16v32double(<32 x i16> %a) {
|
||||
; SSE2: uitofpv32i16v32double
|
||||
; SSE2-LABEL: uitofpv32i16v32double
|
||||
; SSE2: cost of 320 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv32i16v32double
|
||||
; AVX1-LABEL: uitofpv32i16v32double
|
||||
; AVX1: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv32i16v32double
|
||||
; AVX2-LABEL: uitofpv32i16v32double
|
||||
; AVX2: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv32i16v32double
|
||||
; AVX512F-LABEL: uitofpv32i16v32double
|
||||
; AVX512F: cost of 88 {{.*}} uitofp
|
||||
%1 = uitofp <32 x i16> %a to <32 x double>
|
||||
ret <32 x double> %1
|
||||
}
|
||||
|
||||
define <2 x double> @uitofpv2i32v2double(<2 x i32> %a) {
|
||||
; SSE2: uitofpv2i32v2double
|
||||
; SSE2-LABEL: uitofpv2i32v2double
|
||||
; SSE2: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv2i32v2double
|
||||
; AVX1-LABEL: uitofpv2i32v2double
|
||||
; AVX1: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv2i32v2double
|
||||
; AVX2-LABEL: uitofpv2i32v2double
|
||||
; AVX2: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv2i32v2double
|
||||
; AVX512F-LABEL: uitofpv2i32v2double
|
||||
; AVX512F: cost of 4 {{.*}} uitofp
|
||||
%1 = uitofp <2 x i32> %a to <2 x double>
|
||||
ret <2 x double> %1
|
||||
}
|
||||
|
||||
define <4 x double> @uitofpv4i32v4double(<4 x i32> %a) {
|
||||
; SSE2: uitofpv4i32v4double
|
||||
; SSE2-LABEL: uitofpv4i32v4double
|
||||
; SSE2: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv4i32v4double
|
||||
; AVX1-LABEL: uitofpv4i32v4double
|
||||
; AVX1: cost of 6 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv4i32v4double
|
||||
; AVX2-LABEL: uitofpv4i32v4double
|
||||
; AVX2: cost of 6 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv4i32v4double
|
||||
; AVX512F: cost of 6 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv4i32v4double
|
||||
; AVX512F: cost of 1 {{.*}} uitofp
|
||||
%1 = uitofp <4 x i32> %a to <4 x double>
|
||||
ret <4 x double> %1
|
||||
}
|
||||
|
||||
define <8 x double> @uitofpv8i32v8double(<8 x i32> %a) {
|
||||
; SSE2: uitofpv8i32v8double
|
||||
; SSE2-LABEL: uitofpv8i32v8double
|
||||
; SSE2: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv8i32v8double
|
||||
; AVX1-LABEL: uitofpv8i32v8double
|
||||
; AVX1: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv8i32v8double
|
||||
; AVX2-LABEL: uitofpv8i32v8double
|
||||
; AVX2: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv8i32v8double
|
||||
; AVX512F: cost of 22 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv8i32v8double
|
||||
; AVX512F: cost of 1 {{.*}} uitofp
|
||||
%1 = uitofp <8 x i32> %a to <8 x double>
|
||||
ret <8 x double> %1
|
||||
}
|
||||
|
||||
define <16 x double> @uitofpv16i32v16double(<16 x i32> %a) {
|
||||
; SSE2: uitofpv16i32v16double
|
||||
; SSE2-LABEL: uitofpv16i32v16double
|
||||
; SSE2: cost of 160 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv16i32v16double
|
||||
; AVX1-LABEL: uitofpv16i32v16double
|
||||
; AVX1: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv16i32v16double
|
||||
; AVX2-LABEL: uitofpv16i32v16double
|
||||
; AVX2: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv16i32v16double
|
||||
; AVX512F-LABEL: uitofpv16i32v16double
|
||||
; AVX512F: cost of 44 {{.*}} uitofp
|
||||
%1 = uitofp <16 x i32> %a to <16 x double>
|
||||
ret <16 x double> %1
|
||||
}
|
||||
|
||||
define <32 x double> @uitofpv32i32v32double(<32 x i32> %a) {
|
||||
; SSE2: uitofpv32i32v32double
|
||||
; SSE2-LABEL: uitofpv32i32v32double
|
||||
; SSE2: cost of 320 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv32i32v32double
|
||||
; AVX1-LABEL: uitofpv32i32v32double
|
||||
; AVX1: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv32i32v32double
|
||||
; AVX2-LABEL: uitofpv32i32v32double
|
||||
; AVX2: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv32i32v32double
|
||||
; AVX512F-LABEL: uitofpv32i32v32double
|
||||
; AVX512F: cost of 88 {{.*}} uitofp
|
||||
%1 = uitofp <32 x i32> %a to <32 x double>
|
||||
ret <32 x double> %1
|
||||
}
|
||||
|
||||
define <2 x double> @uitofpv2i64v2double(<2 x i64> %a) {
|
||||
; SSE2: uitofpv2i64v2double
|
||||
; SSE2-LABEL: uitofpv2i64v2double
|
||||
; SSE2: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv2i64v2double
|
||||
; AVX1-LABEL: uitofpv2i64v2double
|
||||
; AVX1: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv2i64v2double
|
||||
; AVX2-LABEL: uitofpv2i64v2double
|
||||
; AVX2: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv2i64v2double
|
||||
; AVX512F: cost of 20 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv2i64v2double
|
||||
; AVX512F: cost of 5 {{.*}} uitofp
|
||||
;
|
||||
; AVX512DQ: uitofpv2i64v2double
|
||||
; AVX512DQ: cost of 1 {{.*}} uitofp
|
||||
%1 = uitofp <2 x i64> %a to <2 x double>
|
||||
ret <2 x double> %1
|
||||
}
|
||||
|
||||
define <4 x double> @uitofpv4i64v4double(<4 x i64> %a) {
|
||||
; SSE2: uitofpv4i64v4double
|
||||
; SSE2-LABEL: uitofpv4i64v4double
|
||||
; SSE2: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv4i64v4double
|
||||
; AVX1-LABEL: uitofpv4i64v4double
|
||||
; AVX1: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv4i64v4double
|
||||
; AVX2-LABEL: uitofpv4i64v4double
|
||||
; AVX2: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv4i64v4double
|
||||
; AVX512F: cost of 40 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv4i64v4double
|
||||
; AVX512F: cost of 12 {{.*}} uitofp
|
||||
;
|
||||
; AVX512DQ: uitofpv4i64v4double
|
||||
; AVX512DQ: cost of 1 {{.*}} uitofp
|
||||
%1 = uitofp <4 x i64> %a to <4 x double>
|
||||
ret <4 x double> %1
|
||||
}
|
||||
|
||||
define <8 x double> @uitofpv8i64v8double(<8 x i64> %a) {
|
||||
; SSE2: uitofpv8i64v8double
|
||||
; SSE2-LABEL: uitofpv8i64v8double
|
||||
; SSE2: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv8i64v8double
|
||||
; AVX1-LABEL: uitofpv8i64v8double
|
||||
; AVX1: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv8i64v8double
|
||||
; AVX2-LABEL: uitofpv8i64v8double
|
||||
; AVX2: cost of 20 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv8i64v8double
|
||||
; AVX512F: cost of 22 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv8i64v8double
|
||||
; AVX512F: cost of 26 {{.*}} uitofp
|
||||
;
|
||||
; AVX512DQ: uitofpv8i64v8double
|
||||
; AVX512DQ: cost of 1 {{.*}} uitofp
|
||||
%1 = uitofp <8 x i64> %a to <8 x double>
|
||||
ret <8 x double> %1
|
||||
}
|
||||
|
||||
define <16 x double> @uitofpv16i64v16double(<16 x i64> %a) {
|
||||
; SSE2: uitofpv16i64v16double
|
||||
; SSE2-LABEL: uitofpv16i64v16double
|
||||
; SSE2: cost of 160 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv16i64v16double
|
||||
; AVX1-LABEL: uitofpv16i64v16double
|
||||
; AVX1: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv16i64v16double
|
||||
; AVX2-LABEL: uitofpv16i64v16double
|
||||
; AVX2: cost of 40 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv16i64v16double
|
||||
; AVX512F-LABEL: uitofpv16i64v16double
|
||||
; AVX512F: cost of 44 {{.*}} uitofp
|
||||
;
|
||||
; AVX512DQ: uitofpv16i64v16double
|
||||
; AVX512DQ: cost of 44 {{.*}} uitofp
|
||||
%1 = uitofp <16 x i64> %a to <16 x double>
|
||||
ret <16 x double> %1
|
||||
}
|
||||
|
||||
define <32 x double> @uitofpv32i64v32double(<32 x i64> %a) {
|
||||
; SSE2: uitofpv32i64v32double
|
||||
; SSE2-LABEL: uitofpv32i64v32double
|
||||
; SSE2: cost of 320 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv32i64v32double
|
||||
; AVX1-LABEL: uitofpv32i64v32double
|
||||
; AVX1: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv32i64v32double
|
||||
; AVX2-LABEL: uitofpv32i64v32double
|
||||
; AVX2: cost of 80 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv32i64v32double
|
||||
; AVX512F-LABEL: uitofpv32i64v32double
|
||||
; AVX512F: cost of 88 {{.*}} uitofp
|
||||
;
|
||||
; AVX512DQ: uitofpv32i64v32double
|
||||
; AVX512DQ: cost of 88 {{.*}} uitofp
|
||||
%1 = uitofp <32 x i64> %a to <32 x double>
|
||||
ret <32 x double> %1
|
||||
}
|
||||
|
||||
define <2 x float> @uitofpv2i8v2float(<2 x i8> %a) {
|
||||
; SSE2: uitofpv2i8v2float
|
||||
; SSE2-LABEL: uitofpv2i8v2float
|
||||
; SSE2: cost of 15 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv2i8v2float
|
||||
; AVX1-LABEL: uitofpv2i8v2float
|
||||
; AVX1: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv2i8v2float
|
||||
; AVX2-LABEL: uitofpv2i8v2float
|
||||
; AVX2: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv2i8v2float
|
||||
; AVX512F-LABEL: uitofpv2i8v2float
|
||||
; AVX512F: cost of 4 {{.*}} uitofp
|
||||
%1 = uitofp <2 x i8> %a to <2 x float>
|
||||
ret <2 x float> %1
|
||||
}
|
||||
|
||||
define <4 x float> @uitofpv4i8v4float(<4 x i8> %a) {
|
||||
; SSE2: uitofpv4i8v4float
|
||||
; SSE2-LABEL: uitofpv4i8v4float
|
||||
; SSE2: cost of 8 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv4i8v4float
|
||||
; AVX1-LABEL: uitofpv4i8v4float
|
||||
; AVX1: cost of 2 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv4i8v4float
|
||||
; AVX2-LABEL: uitofpv4i8v4float
|
||||
; AVX2: cost of 2 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv4i8v4float
|
||||
; AVX512F-LABEL: uitofpv4i8v4float
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <4 x i8> %a to <4 x float>
|
||||
ret <4 x float> %1
|
||||
}
|
||||
|
||||
define <8 x float> @uitofpv8i8v8float(<8 x i8> %a) {
|
||||
; SSE2: uitofpv8i8v8float
|
||||
; SSE2-LABEL: uitofpv8i8v8float
|
||||
; SSE2: cost of 15 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv8i8v8float
|
||||
; AVX1-LABEL: uitofpv8i8v8float
|
||||
; AVX1: cost of 5 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv8i8v8float
|
||||
; AVX2-LABEL: uitofpv8i8v8float
|
||||
; AVX2: cost of 5 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv8i8v8float
|
||||
; AVX512F: cost of 5 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv8i8v8float
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <8 x i8> %a to <8 x float>
|
||||
ret <8 x float> %1
|
||||
}
|
||||
|
||||
define <16 x float> @uitofpv16i8v16float(<16 x i8> %a) {
|
||||
; SSE2: uitofpv16i8v16float
|
||||
; SSE2-LABEL: uitofpv16i8v16float
|
||||
; SSE2: cost of 8 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv16i8v16float
|
||||
; AVX1-LABEL: uitofpv16i8v16float
|
||||
; AVX1: cost of 44 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv16i8v16float
|
||||
; AVX2-LABEL: uitofpv16i8v16float
|
||||
; AVX2: cost of 44 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv16i8v16float
|
||||
; AVX512F: cost of 46 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv16i8v16float
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <16 x i8> %a to <16 x float>
|
||||
ret <16 x float> %1
|
||||
}
|
||||
|
||||
define <32 x float> @uitofpv32i8v32float(<32 x i8> %a) {
|
||||
; SSE2: uitofpv32i8v32float
|
||||
; SSE2-LABEL: uitofpv32i8v32float
|
||||
; SSE2: cost of 16 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv32i8v32float
|
||||
; AVX1-LABEL: uitofpv32i8v32float
|
||||
; AVX1: cost of 88 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv32i8v32float
|
||||
; AVX2-LABEL: uitofpv32i8v32float
|
||||
; AVX2: cost of 88 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv32i8v32float
|
||||
; AVX512F-LABEL: uitofpv32i8v32float
|
||||
; AVX512F: cost of 92 {{.*}} uitofp
|
||||
%1 = uitofp <32 x i8> %a to <32 x float>
|
||||
ret <32 x float> %1
|
||||
}
|
||||
|
||||
define <2 x float> @uitofpv2i16v2float(<2 x i16> %a) {
|
||||
; SSE2: uitofpv2i16v2float
|
||||
; SSE2-LABEL: uitofpv2i16v2float
|
||||
; SSE2: cost of 15 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv2i16v2float
|
||||
; AVX1-LABEL: uitofpv2i16v2float
|
||||
; AVX1: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv2i16v2float
|
||||
; AVX2-LABEL: uitofpv2i16v2float
|
||||
; AVX2: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv2i16v2float
|
||||
; AVX512F-LABEL: uitofpv2i16v2float
|
||||
; AVX512F: cost of 4 {{.*}} uitofp
|
||||
%1 = uitofp <2 x i16> %a to <2 x float>
|
||||
ret <2 x float> %1
|
||||
}
|
||||
|
||||
define <4 x float> @uitofpv4i16v4float(<4 x i16> %a) {
|
||||
; SSE2: uitofpv4i16v4float
|
||||
; SSE2-LABEL: uitofpv4i16v4float
|
||||
; SSE2: cost of 8 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv4i16v4float
|
||||
; AVX1-LABEL: uitofpv4i16v4float
|
||||
; AVX1: cost of 2 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv4i16v4float
|
||||
; AVX2-LABEL: uitofpv4i16v4float
|
||||
; AVX2: cost of 2 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv4i16v4float
|
||||
; AVX512F-LABEL: uitofpv4i16v4float
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <4 x i16> %a to <4 x float>
|
||||
ret <4 x float> %1
|
||||
}
|
||||
|
||||
define <8 x float> @uitofpv8i16v8float(<8 x i16> %a) {
|
||||
; SSE2: uitofpv8i16v8float
|
||||
; SSE2-LABEL: uitofpv8i16v8float
|
||||
; SSE2: cost of 15 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv8i16v8float
|
||||
; AVX1-LABEL: uitofpv8i16v8float
|
||||
; AVX1: cost of 5 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv8i16v8float
|
||||
; AVX2-LABEL: uitofpv8i16v8float
|
||||
; AVX2: cost of 5 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv8i16v8float
|
||||
; AVX512F: cost of 5 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv8i16v8float
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <8 x i16> %a to <8 x float>
|
||||
ret <8 x float> %1
|
||||
}
|
||||
|
||||
define <16 x float> @uitofpv16i16v16float(<16 x i16> %a) {
|
||||
; SSE2: uitofpv16i16v16float
|
||||
; SSE2-LABEL: uitofpv16i16v16float
|
||||
; SSE2: cost of 30 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv16i16v16float
|
||||
; AVX1-LABEL: uitofpv16i16v16float
|
||||
; AVX1: cost of 44 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv16i16v16float
|
||||
; AVX2-LABEL: uitofpv16i16v16float
|
||||
; AVX2: cost of 44 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv16i16v16float
|
||||
; AVX512F: cost of 46 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv16i16v16float
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <16 x i16> %a to <16 x float>
|
||||
ret <16 x float> %1
|
||||
}
|
||||
|
||||
define <32 x float> @uitofpv32i16v32float(<32 x i16> %a) {
|
||||
; SSE2: uitofpv32i16v32float
|
||||
; SSE2-LABEL: uitofpv32i16v32float
|
||||
; SSE2: cost of 60 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv32i16v32float
|
||||
; AVX1-LABEL: uitofpv32i16v32float
|
||||
; AVX1: cost of 88 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv32i16v32float
|
||||
; AVX2-LABEL: uitofpv32i16v32float
|
||||
; AVX2: cost of 88 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv32i16v32float
|
||||
; AVX512F-LABEL: uitofpv32i16v32float
|
||||
; AVX512F: cost of 92 {{.*}} uitofp
|
||||
%1 = uitofp <32 x i16> %a to <32 x float>
|
||||
ret <32 x float> %1
|
||||
}
|
||||
|
||||
define <2 x float> @uitofpv2i32v2float(<2 x i32> %a) {
|
||||
; SSE2: uitofpv2i32v2float
|
||||
; SSE2-LABEL: uitofpv2i32v2float
|
||||
; SSE2: cost of 15 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv2i32v2float
|
||||
; AVX1-LABEL: uitofpv2i32v2float
|
||||
; AVX1: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv2i32v2float
|
||||
; AVX2-LABEL: uitofpv2i32v2float
|
||||
; AVX2: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv2i32v2float
|
||||
; AVX512F: cost of 4 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv2i32v2float
|
||||
; AVX512F: cost of 2 {{.*}} uitofp
|
||||
%1 = uitofp <2 x i32> %a to <2 x float>
|
||||
ret <2 x float> %1
|
||||
}
|
||||
|
||||
define <4 x float> @uitofpv4i32v4float(<4 x i32> %a) {
|
||||
; SSE2: uitofpv4i32v4float
|
||||
; SSE2-LABEL: uitofpv4i32v4float
|
||||
; SSE2: cost of 8 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv4i32v4float
|
||||
; AVX1-LABEL: uitofpv4i32v4float
|
||||
; AVX1: cost of 6 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv4i32v4float
|
||||
; AVX2-LABEL: uitofpv4i32v4float
|
||||
; AVX2: cost of 6 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv4i32v4float
|
||||
; AVX512F: cost of 6 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv4i32v4float
|
||||
; AVX512F: cost of 1 {{.*}} uitofp
|
||||
%1 = uitofp <4 x i32> %a to <4 x float>
|
||||
ret <4 x float> %1
|
||||
}
|
||||
|
||||
define <8 x float> @uitofpv8i32v8float(<8 x i32> %a) {
|
||||
; SSE2: uitofpv8i32v8float
|
||||
; SSE2-LABEL: uitofpv8i32v8float
|
||||
; SSE2: cost of 16 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv8i32v8float
|
||||
; AVX1-LABEL: uitofpv8i32v8float
|
||||
; AVX1: cost of 9 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv8i32v8float
|
||||
; AVX2-LABEL: uitofpv8i32v8float
|
||||
; AVX2: cost of 8 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv8i32v8float
|
||||
; AVX512F: cost of 8 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv8i32v8float
|
||||
; AVX512F: cost of 1 {{.*}} uitofp
|
||||
%1 = uitofp <8 x i32> %a to <8 x float>
|
||||
ret <8 x float> %1
|
||||
}
|
||||
|
||||
define <16 x float> @uitofpv16i32v16float(<16 x i32> %a) {
|
||||
; SSE2: uitofpv16i32v16float
|
||||
; SSE2-LABEL: uitofpv16i32v16float
|
||||
; SSE2: cost of 32 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv16i32v16float
|
||||
; AVX1-LABEL: uitofpv16i32v16float
|
||||
; AVX1: cost of 44 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv16i32v16float
|
||||
; AVX2-LABEL: uitofpv16i32v16float
|
||||
; AVX2: cost of 44 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv16i32v16float
|
||||
; AVX512F: cost of 46 {{.*}} uitofp
|
||||
; AVX512F-LABEL: uitofpv16i32v16float
|
||||
; AVX512F: cost of 1 {{.*}} uitofp
|
||||
%1 = uitofp <16 x i32> %a to <16 x float>
|
||||
ret <16 x float> %1
|
||||
}
|
||||
|
||||
define <32 x float> @uitofpv32i32v32float(<32 x i32> %a) {
|
||||
; SSE2: uitofpv32i32v32float
|
||||
; SSE2-LABEL: uitofpv32i32v32float
|
||||
; SSE2: cost of 64 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv32i32v32float
|
||||
; AVX1-LABEL: uitofpv32i32v32float
|
||||
; AVX1: cost of 88 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv32i32v32float
|
||||
; AVX2-LABEL: uitofpv32i32v32float
|
||||
; AVX2: cost of 88 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv32i32v32float
|
||||
; AVX512F-LABEL: uitofpv32i32v32float
|
||||
; AVX512F: cost of 92 {{.*}} uitofp
|
||||
%1 = uitofp <32 x i32> %a to <32 x float>
|
||||
ret <32 x float> %1
|
||||
}
|
||||
|
||||
define <2 x float> @uitofpv2i64v2float(<2 x i64> %a) {
|
||||
; SSE2: uitofpv2i64v2float
|
||||
; SSE2-LABEL: uitofpv2i64v2float
|
||||
; SSE2: cost of 15 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv2i64v2float
|
||||
; AVX1-LABEL: uitofpv2i64v2float
|
||||
; AVX1: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv2i64v2float
|
||||
; AVX2-LABEL: uitofpv2i64v2float
|
||||
; AVX2: cost of 4 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv2i64v2float
|
||||
; AVX512F-LABEL: uitofpv2i64v2float
|
||||
; AVX512F: cost of 4 {{.*}} uitofp
|
||||
%1 = uitofp <2 x i64> %a to <2 x float>
|
||||
ret <2 x float> %1
|
||||
}
|
||||
|
||||
define <4 x float> @uitofpv4i64v4float(<4 x i64> %a) {
|
||||
; SSE2: uitofpv4i64v4float
|
||||
; SSE2-LABEL: uitofpv4i64v4float
|
||||
; SSE2: cost of 30 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv4i64v4float
|
||||
; AVX1-LABEL: uitofpv4i64v4float
|
||||
; AVX1: cost of 10 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv4i64v4float
|
||||
; AVX2-LABEL: uitofpv4i64v4float
|
||||
; AVX2: cost of 10 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv4i64v4float
|
||||
; AVX512F-LABEL: uitofpv4i64v4float
|
||||
; AVX512F: cost of 10 {{.*}} uitofp
|
||||
%1 = uitofp <4 x i64> %a to <4 x float>
|
||||
ret <4 x float> %1
|
||||
}
|
||||
|
||||
define <8 x float> @uitofpv8i64v8float(<8 x i64> %a) {
|
||||
; SSE2: uitofpv8i64v8float
|
||||
; SSE2-LABEL: uitofpv8i64v8float
|
||||
; SSE2: cost of 60 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv8i64v8float
|
||||
; AVX1-LABEL: uitofpv8i64v8float
|
||||
; AVX1: cost of 22 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv8i64v8float
|
||||
; AVX2-LABEL: uitofpv8i64v8float
|
||||
; AVX2: cost of 22 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv8i64v8float
|
||||
; AVX512F-LABEL: uitofpv8i64v8float
|
||||
; AVX512F: cost of 22 {{.*}} uitofp
|
||||
%1 = uitofp <8 x i64> %a to <8 x float>
|
||||
ret <8 x float> %1
|
||||
}
|
||||
|
||||
define <16 x float> @uitofpv16i64v16float(<16 x i64> %a) {
|
||||
; SSE2: uitofpv16i64v16float
|
||||
; SSE2-LABEL: uitofpv16i64v16float
|
||||
; SSE2: cost of 120 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv16i64v16float
|
||||
; AVX1-LABEL: uitofpv16i64v16float
|
||||
; AVX1: cost of 44 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv16i64v16float
|
||||
; AVX2-LABEL: uitofpv16i64v16float
|
||||
; AVX2: cost of 44 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv16i64v16float
|
||||
; AVX512F-LABEL: uitofpv16i64v16float
|
||||
; AVX512F: cost of 46 {{.*}} uitofp
|
||||
%1 = uitofp <16 x i64> %a to <16 x float>
|
||||
ret <16 x float> %1
|
||||
}
|
||||
|
||||
define <32 x float> @uitofpv32i64v32float(<32 x i64> %a) {
|
||||
; SSE2: uitofpv32i64v32float
|
||||
; SSE2-LABEL: uitofpv32i64v32float
|
||||
; SSE2: cost of 240 {{.*}} uitofp
|
||||
;
|
||||
; AVX1: uitofpv32i64v32float
|
||||
; AVX1-LABEL: uitofpv32i64v32float
|
||||
; AVX1: cost of 88 {{.*}} uitofp
|
||||
;
|
||||
; AVX2: uitofpv32i64v32float
|
||||
; AVX2-LABEL: uitofpv32i64v32float
|
||||
; AVX2: cost of 88 {{.*}} uitofp
|
||||
;
|
||||
; AVX512F: uitofpv32i64v32float
|
||||
; AVX512F-LABEL: uitofpv32i64v32float
|
||||
; AVX512F: cost of 92 {{.*}} uitofp
|
||||
%1 = uitofp <32 x i64> %a to <32 x float>
|
||||
ret <32 x float> %1
|
||||
}
|
||||
|
||||
define <8 x i32> @fptouiv8f32v8i32(<8 x float> %a) {
|
||||
; AVX512F-LABEL: fptouiv8f32v8i32
|
||||
; AVX512F: cost of 1 {{.*}} fptoui
|
||||
%1 = fptoui <8 x float> %a to <8 x i32>
|
||||
ret <8 x i32> %1
|
||||
}
|
||||
|
||||
define <4 x i32> @fptouiv4f32v4i32(<4 x float> %a) {
|
||||
; AVX512F-LABEL: fptouiv4f32v4i32
|
||||
; AVX512F: cost of 1 {{.*}} fptoui
|
||||
%1 = fptoui <4 x float> %a to <4 x i32>
|
||||
ret <4 x i32> %1
|
||||
}
|
||||
|
||||
define <2 x i32> @fptouiv2f32v2i32(<2 x float> %a) {
|
||||
; AVX512F-LABEL: fptouiv2f32v2i32
|
||||
; AVX512F: cost of 1 {{.*}} fptoui
|
||||
%1 = fptoui <2 x float> %a to <2 x i32>
|
||||
ret <2 x i32> %1
|
||||
}
|
||||
|
||||
define <16 x i32> @fptouiv16f32v16i32(<16 x float> %a) {
|
||||
; AVX512F-LABEL: fptouiv16f32v16i32
|
||||
; AVX512F: cost of 1 {{.*}} fptoui
|
||||
%1 = fptoui <16 x float> %a to <16 x i32>
|
||||
ret <16 x i32> %1
|
||||
}
|
||||
|
||||
define <8 x i64> @fptouiv8f32v8i64(<8 x float> %a) {
|
||||
; AVX512DQ-LABEL: fptouiv8f32v8i64
|
||||
; AVX512DQ: cost of 1 {{.*}} fptoui
|
||||
%1 = fptoui <8 x float> %a to <8 x i64>
|
||||
ret <8 x i64> %1
|
||||
}
|
||||
|
||||
define <4 x i64> @fptouiv4f32v4i64(<4 x float> %a) {
|
||||
; AVX512DQ-LABEL: fptouiv4f32v4i64
|
||||
; AVX512DQ: cost of 1 {{.*}} fptoui
|
||||
%1 = fptoui <4 x float> %a to <4 x i64>
|
||||
ret <4 x i64> %1
|
||||
}
|
||||
|
||||
define <2 x i64> @fptouiv2f32v2i64(<2 x float> %a) {
|
||||
; AVX512DQ-LABEL: fptouiv2f32v2i64
|
||||
; AVX512DQ: cost of 1 {{.*}} fptoui
|
||||
%1 = fptoui <2 x float> %a to <2 x i64>
|
||||
ret <2 x i64> %1
|
||||
}
|
||||
|
||||
Loading…
Reference in New Issue
Block a user