maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[Alignment][NFC] Migrate TTI::getInterleavedMemoryOpCost to Align 

This is patch is part of a series to introduce an Alignment type.
See this thread for context: http://lists.llvm.org/pipermail/llvm-dev/2019-July/133851.html
See this patch for the introduction of the type: https://reviews.llvm.org/D64790

Differential Revision: https://reviews.llvm.org/D82573
This commit is contained in:
Guillaume Chatelet 2020-06-26 11:00:53 +00:00
parent 048d11de43
commit fdc7c7fb87
18 changed files with 92 additions and 140 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
llvm/include/llvm/Analysis/TargetTransformInfo.h
View File

@ -1053,10 +1053,10 @@ public:
/// \p UseMaskForCond indicates if the memory access is predicated.
/// \p UseMaskForGaps indicates if gaps should be masked.
int getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
unsigned Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput,
bool UseMaskForCond = false, bool UseMaskForGaps = false) const;
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput,
bool UseMaskForCond = false, bool UseMaskForGaps = false) const;
/// Calculate the cost of performing a vector reduction.
///
@ -1435,13 +1435,10 @@ public:
TTI::TargetCostKind CostKind,
const Instruction *I = nullptr) = 0;
virtual int
getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy, unsigned Factor,
ArrayRef<unsigned> Indices, unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind,
bool UseMaskForCond = false,
bool UseMaskForGaps = false) = 0;
virtual int getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond = false, bool UseMaskForGaps = false) = 0;
virtual int getArithmeticReductionCost(unsigned Opcode, VectorType *Ty,
bool IsPairwiseForm,
TTI::TargetCostKind CostKind) = 0;
@ -1857,7 +1854,7 @@ public:
Alignment, CostKind, I);
}
int getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy, unsigned Factor,
ArrayRef<unsigned> Indices, unsigned Alignment,
ArrayRef<unsigned> Indices, Align Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind,
bool UseMaskForCond,

11
llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
View File

@ -478,13 +478,10 @@ public:
return 1;
}
unsigned getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy,
unsigned Factor,
ArrayRef<unsigned> Indices,
unsigned Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind,
bool UseMaskForCond,
bool UseMaskForGaps) {
unsigned getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond, bool UseMaskForGaps) {
return 1;
}

15
llvm/include/llvm/CodeGen/BasicTTIImpl.h
View File

@ -944,13 +944,10 @@ public:
return Cost;
}
unsigned getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy,
unsigned Factor,
ArrayRef<unsigned> Indices,
unsigned Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind,
bool UseMaskForCond = false,
bool UseMaskForGaps = false) {
unsigned getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond = false, bool UseMaskForGaps = false) {
auto *VT = cast<FixedVectorType>(VecTy);
unsigned NumElts = VT->getNumElements();
@ -963,10 +960,10 @@ public:
unsigned Cost;
if (UseMaskForCond || UseMaskForGaps)
Cost = static_cast<T *>(this)->getMaskedMemoryOpCost(
Opcode, VecTy, Align(Alignment), AddressSpace, CostKind);
Opcode, VecTy, Alignment, AddressSpace, CostKind);
else
Cost = static_cast<T *>(this)->getMemoryOpCost(
Opcode, VecTy, Align(Alignment), AddressSpace, CostKind);
Opcode, VecTy, Alignment, AddressSpace, CostKind);
// Legalize the vector type, and get the legalized and unlegalized type
// sizes.

3
llvm/lib/Analysis/TargetTransformInfo.cpp
View File

@ -781,8 +781,7 @@ int TargetTransformInfo::getGatherScatterOpCost(unsigned Opcode, Type *DataTy,
int TargetTransformInfo::getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
unsigned Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond, bool UseMaskForGaps) const {
int Cost = TTIImpl->getInterleavedMemoryOpCost(
Opcode, VecTy, Factor, Indices, Alignment, AddressSpace, CostKind,

2
llvm/lib/CodeGen/InterleavedLoadCombinePass.cpp
View File

@ -1207,7 +1207,7 @@ bool InterleavedLoadCombineImpl::combine(std::list<VectorInfo> &InterleavedLoad,
for (unsigned i = 0; i < Factor; i++)
Indices.push_back(i);
InterleavedCost = TTI.getInterleavedMemoryOpCost(
Instruction::Load, ILTy, Factor, Indices, InsertionPoint->getAlignment(),
Instruction::Load, ILTy, Factor, Indices, InsertionPoint->getAlign(),
InsertionPoint->getPointerAddressSpace());
if (InterleavedCost >= InstructionCost) {

12
llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
View File

@ -725,14 +725,10 @@ int AArch64TTIImpl::getMemoryOpCost(unsigned Opcode, Type *Ty,
return LT.first;
}
int AArch64TTIImpl::getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy,
unsigned Factor,
ArrayRef<unsigned> Indices,
unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind,
bool UseMaskForCond,
bool UseMaskForGaps) {
int AArch64TTIImpl::getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond, bool UseMaskForGaps) {
assert(Factor >= 2 && "Invalid interleave factor");
auto *VecVTy = cast<VectorType>(VecTy);

11
llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
View File

@ -197,12 +197,11 @@ public:
return BaseT::isLegalNTStore(DataType, Alignment);
}
int getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy, unsigned Factor,
ArrayRef<unsigned> Indices, unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false,
bool UseMaskForGaps = false);
int getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false, bool UseMaskForGaps = false);
bool
shouldConsiderAddressTypePromotion(const Instruction &I,

3
llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp
View File

@ -913,8 +913,7 @@ int ARMTTIImpl::getMemoryOpCost(unsigned Opcode, Type *Src,
int ARMTTIImpl::getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
unsigned Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond, bool UseMaskForGaps) {
assert(Factor >= 2 && "Invalid interleave factor");
assert(isa<VectorType>(VecTy) && "Expect a vector type");

11
llvm/lib/Target/ARM/ARMTargetTransformInfo.h
View File

@ -224,12 +224,11 @@ public:
TTI::TargetCostKind CostKind,
const Instruction *I = nullptr);
int getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy, unsigned Factor,
ArrayRef<unsigned> Indices, unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false,
bool UseMaskForGaps = false);
int getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false, bool UseMaskForGaps = false);
unsigned getGatherScatterOpCost(unsigned Opcode, Type *DataTy,
const Value *Ptr, bool VariableMask,

9
llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.cpp
View File

@ -220,11 +220,10 @@ unsigned HexagonTTIImpl::getGatherScatterOpCost(
Alignment, CostKind, I);
}
unsigned HexagonTTIImpl::getInterleavedMemoryOpCost(unsigned Opcode,
Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
unsigned Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind, bool UseMaskForCond,
bool UseMaskForGaps) {
unsigned HexagonTTIImpl::getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond, bool UseMaskForGaps) {
if (Indices.size() != Factor || UseMaskForCond || UseMaskForGaps)
return BaseT::getInterleavedMemoryOpCost(Opcode, VecTy, Factor, Indices,
Alignment, AddressSpace,

10
llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.h
View File

@ -126,11 +126,11 @@ public:
unsigned Alignment,
TTI::TargetCostKind CostKind,
const Instruction *I);
unsigned getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy,
unsigned Factor, ArrayRef<unsigned> Indices, unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false, bool UseMaskForGaps = false);
unsigned getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false, bool UseMaskForGaps = false);
unsigned getCmpSelInstrCost(unsigned Opcode, Type *ValTy, Type *CondTy,
TTI::TargetCostKind CostKind,
const Instruction *I = nullptr);

12
llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
View File

@ -954,14 +954,10 @@ int PPCTTIImpl::getMemoryOpCost(unsigned Opcode, Type *Src,
return Cost;
}
int PPCTTIImpl::getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy,
unsigned Factor,
ArrayRef<unsigned> Indices,
unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind,
bool UseMaskForCond,
bool UseMaskForGaps) {
int PPCTTIImpl::getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond, bool UseMaskForGaps) {
if (UseMaskForCond || UseMaskForGaps)
return BaseT::getInterleavedMemoryOpCost(Opcode, VecTy, Factor, Indices,
Alignment, AddressSpace, CostKind,

13
llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
View File

@ -112,14 +112,11 @@ public:
unsigned AddressSpace,
TTI::TargetCostKind CostKind,
const Instruction *I = nullptr);
int getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy,
unsigned Factor,
ArrayRef<unsigned> Indices,
unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false,
bool UseMaskForGaps = false);
int getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false, bool UseMaskForGaps = false);
unsigned getIntrinsicInstrCost(const IntrinsicCostAttributes &ICA,
TTI::TargetCostKind CostKind);

12
llvm/lib/Target/SystemZ/SystemZTargetTransformInfo.cpp
View File

@ -1102,14 +1102,10 @@ int SystemZTTIImpl::getMemoryOpCost(unsigned Opcode, Type *Src,
// needed for using / defining the vector operands. The SystemZ version does
// roughly the same but bases the computations on vector permutations
// instead.
int SystemZTTIImpl::getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy,
unsigned Factor,
ArrayRef<unsigned> Indices,
unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind,
bool UseMaskForCond,
bool UseMaskForGaps) {
int SystemZTTIImpl::getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond, bool UseMaskForGaps) {
if (UseMaskForCond || UseMaskForGaps)
return BaseT::getInterleavedMemoryOpCost(Opcode, VecTy, Factor, Indices,
Alignment, AddressSpace, CostKind,

13
llvm/lib/Target/SystemZ/SystemZTargetTransformInfo.h
View File

@ -101,14 +101,11 @@ public:
unsigned AddressSpace, TTI::TargetCostKind CostKind,
const Instruction *I = nullptr);
int getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy,
unsigned Factor,
ArrayRef<unsigned> Indices,
unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false,
bool UseMaskForGaps = false);
int getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false, bool UseMaskForGaps = false);
int getIntrinsicInstrCost(const IntrinsicCostAttributes &ICA,
TTI::TargetCostKind CostKind);

36
llvm/lib/Target/X86/X86TargetTransformInfo.cpp
View File

@ -4244,14 +4244,10 @@ bool X86TTIImpl::enableInterleavedAccessVectorization() {
// computing the cost using a generic formula as a function of generic
// shuffles. We therefore use a lookup table instead, filled according to
// the instruction sequences that codegen currently generates.
int X86TTIImpl::getInterleavedMemoryOpCostAVX2(unsigned Opcode, Type *VecTy,
unsigned Factor,
ArrayRef<unsigned> Indices,
unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind,
bool UseMaskForCond,
bool UseMaskForGaps) {
int X86TTIImpl::getInterleavedMemoryOpCostAVX2(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond, bool UseMaskForGaps) {
if (UseMaskForCond || UseMaskForGaps)
return BaseT::getInterleavedMemoryOpCost(Opcode, VecTy, Factor, Indices,
@ -4366,14 +4362,10 @@ int X86TTIImpl::getInterleavedMemoryOpCostAVX2(unsigned Opcode, Type *VecTy,
// \p Indices contains indices for strided load.
// \p Factor - the factor of interleaving.
// AVX-512 provides 3-src shuffles that significantly reduces the cost.
int X86TTIImpl::getInterleavedMemoryOpCostAVX512(unsigned Opcode, Type *VecTy,
unsigned Factor,
ArrayRef<unsigned> Indices,
unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind,
bool UseMaskForCond,
bool UseMaskForGaps) {
int X86TTIImpl::getInterleavedMemoryOpCostAVX512(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond, bool UseMaskForGaps) {
if (UseMaskForCond || UseMaskForGaps)
return BaseT::getInterleavedMemoryOpCost(Opcode, VecTy, Factor, Indices,
@ -4495,14 +4487,10 @@ int X86TTIImpl::getInterleavedMemoryOpCostAVX512(unsigned Opcode, Type *VecTy,
return Cost;
}
int X86TTIImpl::getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy,
unsigned Factor,
ArrayRef<unsigned> Indices,
unsigned Alignment,
unsigned AddressSpace,
TTI::TargetCostKind CostKind,
bool UseMaskForCond,
bool UseMaskForGaps) {
int X86TTIImpl::getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace, TTI::TargetCostKind CostKind,
bool UseMaskForCond, bool UseMaskForGaps) {
auto isSupportedOnAVX512 = [](Type *VecTy, bool HasBW) {
Type *EltTy = cast<VectorType>(VecTy)->getElementType();
if (EltTy->isFloatTy() || EltTy->isDoubleTy() || EltTy->isIntegerTy(64) ||

33
llvm/lib/Target/X86/X86TargetTransformInfo.h
View File

@ -168,24 +168,21 @@ public:
bool IsPairwiseForm, bool IsUnsigned,
TTI::TargetCostKind CostKind);
int getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy,
unsigned Factor, ArrayRef<unsigned> Indices,
unsigned Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false,
bool UseMaskForGaps = false);
int getInterleavedMemoryOpCostAVX512(unsigned Opcode, Type *VecTy,
unsigned Factor, ArrayRef<unsigned> Indices,
unsigned Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false,
bool UseMaskForGaps = false);
int getInterleavedMemoryOpCostAVX2(unsigned Opcode, Type *VecTy,
unsigned Factor, ArrayRef<unsigned> Indices,
unsigned Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false,
bool UseMaskForGaps = false);
int getInterleavedMemoryOpCost(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false, bool UseMaskForGaps = false);
int getInterleavedMemoryOpCostAVX512(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false, bool UseMaskForGaps = false);
int getInterleavedMemoryOpCostAVX2(
unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
Align Alignment, unsigned AddressSpace,
TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
bool UseMaskForCond = false, bool UseMaskForGaps = false);
int getIntImmCost(int64_t);

5
llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
View File

@ -5966,9 +5966,8 @@ unsigned LoopVectorizationCostModel::getInterleaveGroupCost(Instruction *I,
bool UseMaskForGaps =
Group->requiresScalarEpilogue() && !isScalarEpilogueAllowed();
unsigned Cost = TTI.getInterleavedMemoryOpCost(
I->getOpcode(), WideVecTy, Group->getFactor(), Indices,
Group->getAlign().value(), AS, TTI::TCK_RecipThroughput,
Legal->isMaskRequired(I), UseMaskForGaps);
I->getOpcode(), WideVecTy, Group->getFactor(), Indices, Group->getAlign(),
AS, TTI::TCK_RecipThroughput, Legal->isMaskRequired(I), UseMaskForGaps);
if (Group->isReverse()) {
// TODO: Add support for reversed masked interleaved access.