namespace {
-class BasicTTI : public ImmutablePass, public TargetTransformInfo {
- const TargetLowering *TLI;
+class BasicTTI LLVM_FINAL : public ImmutablePass, public TargetTransformInfo {
+ const TargetMachine *TM;
/// Estimate the overhead of scalarizing an instruction. Insert and Extract
/// are set if the result needs to be inserted and/or extracted from vectors.
unsigned getScalarizationOverhead(Type *Ty, bool Insert, bool Extract) const;
+ const TargetLoweringBase *getTLI() const { return TM->getTargetLowering(); }
+
public:
- BasicTTI() : ImmutablePass(ID), TLI(0) {
+ BasicTTI() : ImmutablePass(ID), TM(0) {
llvm_unreachable("This pass cannot be directly constructed");
}
- BasicTTI(const TargetLowering *TLI) : ImmutablePass(ID), TLI(TLI) {
+ BasicTTI(const TargetMachine *TM) : ImmutablePass(ID), TM(TM) {
initializeBasicTTIPass(*PassRegistry::getPassRegistry());
}
- virtual void initializePass() {
+ virtual void initializePass() LLVM_OVERRIDE {
pushTTIStack(this);
}
popTTIStack();
}
- virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+ virtual void getAnalysisUsage(AnalysisUsage &AU) const LLVM_OVERRIDE {
TargetTransformInfo::getAnalysisUsage(AU);
}
static char ID;
/// Provide necessary pointer adjustments for the two base classes.
- virtual void *getAdjustedAnalysisPointer(const void *ID) {
+ virtual void *getAdjustedAnalysisPointer(const void *ID) LLVM_OVERRIDE {
if (ID == &TargetTransformInfo::ID)
return (TargetTransformInfo*)this;
return this;
}
+ virtual bool hasBranchDivergence() const LLVM_OVERRIDE;
+
/// \name Scalar TTI Implementations
/// @{
- virtual bool isLegalAddImmediate(int64_t imm) const;
- virtual bool isLegalICmpImmediate(int64_t imm) const;
+ virtual bool isLegalAddImmediate(int64_t imm) const LLVM_OVERRIDE;
+ virtual bool isLegalICmpImmediate(int64_t imm) const LLVM_OVERRIDE;
virtual bool isLegalAddressingMode(Type *Ty, GlobalValue *BaseGV,
int64_t BaseOffset, bool HasBaseReg,
- int64_t Scale) const;
- virtual bool isTruncateFree(Type *Ty1, Type *Ty2) const;
- virtual bool isTypeLegal(Type *Ty) const;
- virtual unsigned getJumpBufAlignment() const;
- virtual unsigned getJumpBufSize() const;
- virtual bool shouldBuildLookupTables() const;
+ int64_t Scale) const LLVM_OVERRIDE;
+ virtual int getScalingFactorCost(Type *Ty, GlobalValue *BaseGV,
+ int64_t BaseOffset, bool HasBaseReg,
+ int64_t Scale) const LLVM_OVERRIDE;
+ virtual bool isTruncateFree(Type *Ty1, Type *Ty2) const LLVM_OVERRIDE;
+ virtual bool isTypeLegal(Type *Ty) const LLVM_OVERRIDE;
+ virtual unsigned getJumpBufAlignment() const LLVM_OVERRIDE;
+ virtual unsigned getJumpBufSize() const LLVM_OVERRIDE;
+ virtual bool shouldBuildLookupTables() const LLVM_OVERRIDE;
+ virtual bool haveFastSqrt(Type *Ty) const LLVM_OVERRIDE;
+ virtual void getUnrollingPreferences(
+ Loop *L, UnrollingPreferences &UP) const LLVM_OVERRIDE;
/// @}
/// \name Vector TTI Implementations
/// @{
- virtual unsigned getNumberOfRegisters(bool Vector) const;
- virtual unsigned getMaximumUnrollFactor() const;
- virtual unsigned getRegisterBitWidth(bool Vector) const;
- virtual unsigned getArithmeticInstrCost(unsigned Opcode, Type *Ty) const;
+ virtual unsigned getNumberOfRegisters(bool Vector) const LLVM_OVERRIDE;
+ virtual unsigned getMaximumUnrollFactor() const LLVM_OVERRIDE;
+ virtual unsigned getRegisterBitWidth(bool Vector) const LLVM_OVERRIDE;
+ virtual unsigned getArithmeticInstrCost(unsigned Opcode, Type *Ty,
+ OperandValueKind,
+ OperandValueKind) const LLVM_OVERRIDE;
virtual unsigned getShuffleCost(ShuffleKind Kind, Type *Tp,
- int Index, Type *SubTp) const;
+ int Index, Type *SubTp) const LLVM_OVERRIDE;
virtual unsigned getCastInstrCost(unsigned Opcode, Type *Dst,
- Type *Src) const;
- virtual unsigned getCFInstrCost(unsigned Opcode) const;
+ Type *Src) const LLVM_OVERRIDE;
+ virtual unsigned getCFInstrCost(unsigned Opcode) const LLVM_OVERRIDE;
virtual unsigned getCmpSelInstrCost(unsigned Opcode, Type *ValTy,
- Type *CondTy) const;
+ Type *CondTy) const LLVM_OVERRIDE;
virtual unsigned getVectorInstrCost(unsigned Opcode, Type *Val,
- unsigned Index) const;
+ unsigned Index) const LLVM_OVERRIDE;
virtual unsigned getMemoryOpCost(unsigned Opcode, Type *Src,
unsigned Alignment,
- unsigned AddressSpace) const;
- virtual unsigned getIntrinsicInstrCost(Intrinsic::ID, Type *RetTy,
- ArrayRef<Type*> Tys) const;
- virtual unsigned getNumberOfParts(Type *Tp) const;
+ unsigned AddressSpace) const LLVM_OVERRIDE;
+ virtual unsigned getIntrinsicInstrCost(
+ Intrinsic::ID, Type *RetTy, ArrayRef<Type*> Tys) const LLVM_OVERRIDE;
+ virtual unsigned getNumberOfParts(Type *Tp) const LLVM_OVERRIDE;
+ virtual unsigned getAddressComputationCost(
+ Type *Ty, bool IsComplex) const LLVM_OVERRIDE;
+ virtual unsigned getReductionCost(unsigned Opcode, Type *Ty,
+ bool IsPairwise) const LLVM_OVERRIDE;
/// @}
};
char BasicTTI::ID = 0;
ImmutablePass *
-llvm::createBasicTargetTransformInfoPass(const TargetLowering *TLI) {
- return new BasicTTI(TLI);
+llvm::createBasicTargetTransformInfoPass(const TargetMachine *TM) {
+ return new BasicTTI(TM);
}
+bool BasicTTI::hasBranchDivergence() const { return false; }
bool BasicTTI::isLegalAddImmediate(int64_t imm) const {
- return TLI->isLegalAddImmediate(imm);
+ return getTLI()->isLegalAddImmediate(imm);
}
bool BasicTTI::isLegalICmpImmediate(int64_t imm) const {
- return TLI->isLegalICmpImmediate(imm);
+ return getTLI()->isLegalICmpImmediate(imm);
}
bool BasicTTI::isLegalAddressingMode(Type *Ty, GlobalValue *BaseGV,
int64_t BaseOffset, bool HasBaseReg,
int64_t Scale) const {
- TargetLowering::AddrMode AM;
+ TargetLoweringBase::AddrMode AM;
AM.BaseGV = BaseGV;
AM.BaseOffs = BaseOffset;
AM.HasBaseReg = HasBaseReg;
AM.Scale = Scale;
- return TLI->isLegalAddressingMode(AM, Ty);
+ return getTLI()->isLegalAddressingMode(AM, Ty);
+}
+
+int BasicTTI::getScalingFactorCost(Type *Ty, GlobalValue *BaseGV,
+ int64_t BaseOffset, bool HasBaseReg,
+ int64_t Scale) const {
+ TargetLoweringBase::AddrMode AM;
+ AM.BaseGV = BaseGV;
+ AM.BaseOffs = BaseOffset;
+ AM.HasBaseReg = HasBaseReg;
+ AM.Scale = Scale;
+ return getTLI()->getScalingFactorCost(AM, Ty);
}
bool BasicTTI::isTruncateFree(Type *Ty1, Type *Ty2) const {
- return TLI->isTruncateFree(Ty1, Ty2);
+ return getTLI()->isTruncateFree(Ty1, Ty2);
}
bool BasicTTI::isTypeLegal(Type *Ty) const {
- EVT T = TLI->getValueType(Ty);
- return TLI->isTypeLegal(T);
+ EVT T = getTLI()->getValueType(Ty);
+ return getTLI()->isTypeLegal(T);
}
unsigned BasicTTI::getJumpBufAlignment() const {
- return TLI->getJumpBufAlignment();
+ return getTLI()->getJumpBufAlignment();
}
unsigned BasicTTI::getJumpBufSize() const {
- return TLI->getJumpBufSize();
+ return getTLI()->getJumpBufSize();
}
bool BasicTTI::shouldBuildLookupTables() const {
+ const TargetLoweringBase *TLI = getTLI();
return TLI->supportJumpTables() &&
(TLI->isOperationLegalOrCustom(ISD::BR_JT, MVT::Other) ||
TLI->isOperationLegalOrCustom(ISD::BRIND, MVT::Other));
}
+bool BasicTTI::haveFastSqrt(Type *Ty) const {
+ const TargetLoweringBase *TLI = getTLI();
+ EVT VT = TLI->getValueType(Ty);
+ return TLI->isTypeLegal(VT) && TLI->isOperationLegalOrCustom(ISD::FSQRT, VT);
+}
+
+void BasicTTI::getUnrollingPreferences(Loop *, UnrollingPreferences &) const { }
+
//===----------------------------------------------------------------------===//
//
// Calls used by the vectorizers.
return 1;
}
-unsigned BasicTTI::getArithmeticInstrCost(unsigned Opcode, Type *Ty) const {
+unsigned BasicTTI::getArithmeticInstrCost(unsigned Opcode, Type *Ty,
+ OperandValueKind,
+ OperandValueKind) const {
// Check if any of the operands are vector operands.
+ const TargetLoweringBase *TLI = getTLI();
int ISD = TLI->InstructionOpcodeToISD(Opcode);
assert(ISD && "Invalid opcode");
std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Ty);
+ bool IsFloat = Ty->getScalarType()->isFloatingPointTy();
+ // Assume that floating point arithmetic operations cost twice as much as
+ // integer operations.
+ unsigned OpCost = (IsFloat ? 2 : 1);
+
if (TLI->isOperationLegalOrPromote(ISD, LT.second)) {
// The operation is legal. Assume it costs 1.
- // If the type is split to multiple registers, assume that thre is some
+ // If the type is split to multiple registers, assume that there is some
// overhead to this.
// TODO: Once we have extract/insert subvector cost we need to use them.
if (LT.first > 1)
- return LT.first * 2;
- return LT.first * 1;
+ return LT.first * 2 * OpCost;
+ return LT.first * 1 * OpCost;
}
if (!TLI->isOperationExpand(ISD, LT.second)) {
// If the operation is custom lowered then assume
// thare the code is twice as expensive.
- return LT.first * 2;
+ return LT.first * 2 * OpCost;
}
// Else, assume that we need to scalarize this op.
}
// We don't know anything about this scalar instruction.
- return 1;
+ return OpCost;
}
unsigned BasicTTI::getShuffleCost(ShuffleKind Kind, Type *Tp, int Index,
unsigned BasicTTI::getCastInstrCost(unsigned Opcode, Type *Dst,
Type *Src) const {
+ const TargetLoweringBase *TLI = getTLI();
int ISD = TLI->InstructionOpcodeToISD(Opcode);
assert(ISD && "Invalid opcode");
std::pair<unsigned, MVT> SrcLT = TLI->getTypeLegalizationCost(Src);
std::pair<unsigned, MVT> DstLT = TLI->getTypeLegalizationCost(Dst);
+ // Check for NOOP conversions.
+ if (SrcLT.first == DstLT.first &&
+ SrcLT.second.getSizeInBits() == DstLT.second.getSizeInBits()) {
+
+ // Bitcast between types that are legalized to the same type are free.
+ if (Opcode == Instruction::BitCast || Opcode == Instruction::Trunc)
+ return 0;
+ }
+
+ if (Opcode == Instruction::Trunc &&
+ TLI->isTruncateFree(SrcLT.second, DstLT.second))
+ return 0;
+
+ if (Opcode == Instruction::ZExt &&
+ TLI->isZExtFree(SrcLT.second, DstLT.second))
+ return 0;
+
+ // If the cast is marked as legal (or promote) then assume low cost.
+ if (TLI->isOperationLegalOrPromote(ISD, DstLT.second))
+ return 1;
+
// Handle scalar conversions.
if (!Src->isVectorTy() && !Dst->isVectorTy()) {
if (Opcode == Instruction::BitCast)
return 0;
- if (Opcode == Instruction::Trunc &&
- TLI->isTruncateFree(SrcLT.second, DstLT.second))
- return 0;
-
- if (Opcode == Instruction::ZExt &&
- TLI->isZExtFree(SrcLT.second, DstLT.second))
- return 0;
-
// Just check the op cost. If the operation is legal then assume it costs 1.
if (!TLI->isOperationExpand(ISD, DstLT.second))
return 1;
if (SrcLT.first == DstLT.first &&
SrcLT.second.getSizeInBits() == DstLT.second.getSizeInBits()) {
- // Bitcast between types that are legalized to the same type are free.
- if (Opcode == Instruction::BitCast || Opcode == Instruction::Trunc)
- return 0;
-
// Assume that Zext is done using AND.
if (Opcode == Instruction::ZExt)
return 1;
unsigned BasicTTI::getCmpSelInstrCost(unsigned Opcode, Type *ValTy,
Type *CondTy) const {
+ const TargetLoweringBase *TLI = getTLI();
int ISD = TLI->InstructionOpcodeToISD(Opcode);
assert(ISD && "Invalid opcode");
unsigned Alignment,
unsigned AddressSpace) const {
assert(!Src->isVoidTy() && "Invalid type");
- std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Src);
+ std::pair<unsigned, MVT> LT = getTLI()->getTypeLegalizationCost(Src);
// Assume that all loads of legal types cost 1.
return LT.first;
}
-unsigned BasicTTI::getIntrinsicInstrCost(Intrinsic::ID, Type *RetTy,
+unsigned BasicTTI::getIntrinsicInstrCost(Intrinsic::ID IID, Type *RetTy,
ArrayRef<Type *> Tys) const {
- // assume that we need to scalarize this intrinsic.
- unsigned ScalarizationCost = 0;
- unsigned ScalarCalls = 1;
- if (RetTy->isVectorTy()) {
- ScalarizationCost = getScalarizationOverhead(RetTy, true, false);
- ScalarCalls = std::max(ScalarCalls, RetTy->getVectorNumElements());
- }
- for (unsigned i = 0, ie = Tys.size(); i != ie; ++i) {
- if (Tys[i]->isVectorTy()) {
- ScalarizationCost += getScalarizationOverhead(Tys[i], false, true);
+ unsigned ISD = 0;
+ switch (IID) {
+ default: {
+ // Assume that we need to scalarize this intrinsic.
+ unsigned ScalarizationCost = 0;
+ unsigned ScalarCalls = 1;
+ if (RetTy->isVectorTy()) {
+ ScalarizationCost = getScalarizationOverhead(RetTy, true, false);
ScalarCalls = std::max(ScalarCalls, RetTy->getVectorNumElements());
}
+ for (unsigned i = 0, ie = Tys.size(); i != ie; ++i) {
+ if (Tys[i]->isVectorTy()) {
+ ScalarizationCost += getScalarizationOverhead(Tys[i], false, true);
+ ScalarCalls = std::max(ScalarCalls, RetTy->getVectorNumElements());
+ }
+ }
+
+ return ScalarCalls + ScalarizationCost;
+ }
+ // Look for intrinsics that can be lowered directly or turned into a scalar
+ // intrinsic call.
+ case Intrinsic::sqrt: ISD = ISD::FSQRT; break;
+ case Intrinsic::sin: ISD = ISD::FSIN; break;
+ case Intrinsic::cos: ISD = ISD::FCOS; break;
+ case Intrinsic::exp: ISD = ISD::FEXP; break;
+ case Intrinsic::exp2: ISD = ISD::FEXP2; break;
+ case Intrinsic::log: ISD = ISD::FLOG; break;
+ case Intrinsic::log10: ISD = ISD::FLOG10; break;
+ case Intrinsic::log2: ISD = ISD::FLOG2; break;
+ case Intrinsic::fabs: ISD = ISD::FABS; break;
+ case Intrinsic::copysign: ISD = ISD::FCOPYSIGN; break;
+ case Intrinsic::floor: ISD = ISD::FFLOOR; break;
+ case Intrinsic::ceil: ISD = ISD::FCEIL; break;
+ case Intrinsic::trunc: ISD = ISD::FTRUNC; break;
+ case Intrinsic::nearbyint:
+ ISD = ISD::FNEARBYINT; break;
+ case Intrinsic::rint: ISD = ISD::FRINT; break;
+ case Intrinsic::round: ISD = ISD::FROUND; break;
+ case Intrinsic::pow: ISD = ISD::FPOW; break;
+ case Intrinsic::fma: ISD = ISD::FMA; break;
+ case Intrinsic::fmuladd: ISD = ISD::FMA; break; // FIXME: mul + add?
+ case Intrinsic::lifetime_start:
+ case Intrinsic::lifetime_end:
+ return 0;
+ }
+
+ const TargetLoweringBase *TLI = getTLI();
+ std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(RetTy);
+
+ if (TLI->isOperationLegalOrPromote(ISD, LT.second)) {
+ // The operation is legal. Assume it costs 1.
+ // If the type is split to multiple registers, assume that thre is some
+ // overhead to this.
+ // TODO: Once we have extract/insert subvector cost we need to use them.
+ if (LT.first > 1)
+ return LT.first * 2;
+ return LT.first * 1;
+ }
+
+ if (!TLI->isOperationExpand(ISD, LT.second)) {
+ // If the operation is custom lowered then assume
+ // thare the code is twice as expensive.
+ return LT.first * 2;
+ }
+
+ // Else, assume that we need to scalarize this intrinsic. For math builtins
+ // this will emit a costly libcall, adding call overhead and spills. Make it
+ // very expensive.
+ if (RetTy->isVectorTy()) {
+ unsigned Num = RetTy->getVectorNumElements();
+ unsigned Cost = TopTTI->getIntrinsicInstrCost(IID, RetTy->getScalarType(),
+ Tys);
+ return 10 * Cost * Num;
}
- return ScalarCalls + ScalarizationCost;
+
+ // This is going to be turned into a library call, make it expensive.
+ return 10;
}
unsigned BasicTTI::getNumberOfParts(Type *Tp) const {
- std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Tp);
+ std::pair<unsigned, MVT> LT = getTLI()->getTypeLegalizationCost(Tp);
return LT.first;
}
+
+unsigned BasicTTI::getAddressComputationCost(Type *Ty, bool IsComplex) const {
+ return 0;
+}
+
+unsigned BasicTTI::getReductionCost(unsigned Opcode, Type *Ty,
+ bool IsPairwise) const {
+ assert(Ty->isVectorTy() && "Expect a vector type");
+ unsigned NumVecElts = Ty->getVectorNumElements();
+ unsigned NumReduxLevels = Log2_32(NumVecElts);
+ unsigned ArithCost = NumReduxLevels *
+ TopTTI->getArithmeticInstrCost(Opcode, Ty);
+ // Assume the pairwise shuffles add a cost.
+ unsigned ShuffleCost =
+ NumReduxLevels * (IsPairwise + 1) *
+ TopTTI->getShuffleCost(SK_ExtractSubvector, Ty, NumVecElts / 2, Ty);
+ return ShuffleCost + ArithCost + getScalarizationOverhead(Ty, false, true);
+}
projects / oota-llvm.git / blobdiff