diff options
Diffstat (limited to 'llvm/lib/Target/X86/X86TargetTransformInfo.cpp')
| -rw-r--r-- | llvm/lib/Target/X86/X86TargetTransformInfo.cpp | 32 |
1 files changed, 14 insertions, 18 deletions
diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp index 05e5424..fa3f97e 100644 --- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp +++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp @@ -173,14 +173,12 @@ unsigned X86TTIImpl::getMaxInterleaveFactor(unsigned VF) { return 2; } -int X86TTIImpl::getArithmeticInstrCost(unsigned Opcode, Type *Ty, - TTI::TargetCostKind CostKind, - TTI::OperandValueKind Op1Info, - TTI::OperandValueKind Op2Info, - TTI::OperandValueProperties Opd1PropInfo, - TTI::OperandValueProperties Opd2PropInfo, - ArrayRef<const Value *> Args, - const Instruction *CxtI) { +InstructionCost X86TTIImpl::getArithmeticInstrCost( + unsigned Opcode, Type *Ty, TTI::TargetCostKind CostKind, + TTI::OperandValueKind Op1Info, TTI::OperandValueKind Op2Info, + TTI::OperandValueProperties Opd1PropInfo, + TTI::OperandValueProperties Opd2PropInfo, ArrayRef<const Value *> Args, + const Instruction *CxtI) { // TODO: Handle more cost kinds. if (CostKind != TTI::TCK_RecipThroughput) return BaseT::getArithmeticInstrCost(Opcode, Ty, CostKind, Op1Info, @@ -265,10 +263,9 @@ int X86TTIImpl::getArithmeticInstrCost(unsigned Opcode, Type *Ty, // normally expanded to the sequence SRA + SRL + ADD + SRA. // The OperandValue properties may not be the same as that of the previous // operation; conservatively assume OP_None. - int Cost = + InstructionCost Cost = 2 * getArithmeticInstrCost(Instruction::AShr, Ty, CostKind, Op1Info, - Op2Info, - TargetTransformInfo::OP_None, + Op2Info, TargetTransformInfo::OP_None, TargetTransformInfo::OP_None); Cost += getArithmeticInstrCost(Instruction::LShr, Ty, CostKind, Op1Info, Op2Info, @@ -953,7 +950,7 @@ int X86TTIImpl::getArithmeticInstrCost(unsigned Opcode, Type *Ty, // to hide "20 cycles" for each lane. if (LT.second.isVector() && (ISD == ISD::SDIV || ISD == ISD::SREM || ISD == ISD::UDIV || ISD == ISD::UREM)) { - int ScalarCost = getArithmeticInstrCost( + InstructionCost ScalarCost = getArithmeticInstrCost( Opcode, Ty->getScalarType(), CostKind, Op1Info, Op2Info, TargetTransformInfo::OP_None, TargetTransformInfo::OP_None); return 20 * LT.first * LT.second.getVectorNumElements() * ScalarCost; @@ -3419,14 +3416,13 @@ X86TTIImpl::getArithmeticReductionCost(unsigned Opcode, VectorType *ValTy, if (ISD == ISD::MUL && MTy.getScalarType() == MVT::i8) { auto *WideSclTy = IntegerType::get(ValVTy->getContext(), 16); auto *WideVecTy = FixedVectorType::get(WideSclTy, ValVTy->getNumElements()); - return *getCastInstrCost(Instruction::ZExt, WideVecTy, ValTy, - TargetTransformInfo::CastContextHint::None, - CostKind) - .getValue() + + return getCastInstrCost(Instruction::ZExt, WideVecTy, ValTy, + TargetTransformInfo::CastContextHint::None, + CostKind) + getArithmeticReductionCost(Opcode, WideVecTy, IsPairwise, CostKind); } - unsigned ArithmeticCost = 0; + InstructionCost ArithmeticCost = 0; if (LT.first != 1 && MTy.isVector() && MTy.getVectorNumElements() < ValVTy->getNumElements()) { // Type needs to be split. We need LT.first - 1 arithmetic ops. @@ -3496,7 +3492,7 @@ X86TTIImpl::getArithmeticReductionCost(unsigned Opcode, VectorType *ValTy, // Handle bool allof/anyof patterns. if (ValVTy->getElementType()->isIntegerTy(1)) { - unsigned ArithmeticCost = 0; + InstructionCost ArithmeticCost = 0; if (LT.first != 1 && MTy.isVector() && MTy.getVectorNumElements() < ValVTy->getNumElements()) { // Type needs to be split. We need LT.first - 1 arithmetic ops. |