}
bool ScalarTargetTransformImpl::isLegalAddressingMode(const AddrMode &AM,
- Type *Ty) const {
+ Type *Ty) const {
return TLI->isLegalAddressingMode(AM, Ty);
}
return TLI->getJumpBufSize();
}
+bool ScalarTargetTransformImpl::shouldBuildLookupTables() const {
+ return TLI->supportJumpTables() &&
+ (TLI->isOperationLegalOrCustom(ISD::BR_JT, MVT::Other) ||
+ TLI->isOperationLegalOrCustom(ISD::BRIND, MVT::Other));
+}
+
//===----------------------------------------------------------------------===//
//
// Calls used by the vectorizers.
//
//===----------------------------------------------------------------------===//
-static int InstructionOpcodeToISD(unsigned Opcode) {
+int VectorTargetTransformImpl::InstructionOpcodeToISD(unsigned Opcode) const {
enum InstructionOpcodes {
#define HANDLE_INST(NUM, OPCODE, CLASS) OPCODE = NUM,
#define LAST_OTHER_INST(NUM) InstructionOpcodesCount = NUM
case AtomicRMW: return 0;
case Trunc: return ISD::TRUNCATE;
case ZExt: return ISD::ZERO_EXTEND;
- case SExt: return ISD::SEXTLOAD;
+ case SExt: return ISD::SIGN_EXTEND;
case FPToUI: return ISD::FP_TO_UINT;
case FPToSI: return ISD::FP_TO_SINT;
case UIToFP: return ISD::UINT_TO_FP;
llvm_unreachable("Unknown instruction type encountered!");
}
-std::pair<unsigned, EVT>
-VectorTargetTransformImpl::getTypeLegalizationCost(LLVMContext &C,
- EVT Ty) const {
+std::pair<unsigned, MVT>
+VectorTargetTransformImpl::getTypeLegalizationCost(Type *Ty) const {
+ LLVMContext &C = Ty->getContext();
+ EVT MTy = TLI->getValueType(Ty);
+
unsigned Cost = 1;
// We keep legalizing the type until we find a legal kind. We assume that
// the only operation that costs anything is the split. After splitting
// we need to handle two types.
while (true) {
- TargetLowering::LegalizeKind LK = TLI->getTypeConversion(C, Ty);
+ TargetLowering::LegalizeKind LK = TLI->getTypeConversion(C, MTy);
if (LK.first == TargetLowering::TypeLegal)
- return std::make_pair(Cost, LK.second);
+ return std::make_pair(Cost, MTy.getSimpleVT());
- if (LK.first == TargetLowering::TypeSplitVector)
+ if (LK.first == TargetLowering::TypeSplitVector ||
+ LK.first == TargetLowering::TypeExpandInteger)
Cost *= 2;
// Keep legalizing the type.
- Ty = LK.second;
+ MTy = LK.second;
}
}
unsigned
-VectorTargetTransformImpl::getInstrCost(unsigned Opcode, Type *Ty1,
- Type *Ty2) const {
+VectorTargetTransformImpl::getScalarizationOverhead(Type *Ty,
+ bool Insert,
+ bool Extract) const {
+ assert (Ty->isVectorTy() && "Can only scalarize vectors");
+ unsigned Cost = 0;
+
+ for (int i = 0, e = Ty->getVectorNumElements(); i < e; ++i) {
+ if (Insert)
+ Cost += getVectorInstrCost(Instruction::InsertElement, Ty, i);
+ if (Extract)
+ Cost += getVectorInstrCost(Instruction::ExtractElement, Ty, i);
+ }
+
+ return Cost;
+}
+
+unsigned VectorTargetTransformImpl::getArithmeticInstrCost(unsigned Opcode,
+ Type *Ty) const {
// Check if any of the operands are vector operands.
int ISD = InstructionOpcodeToISD(Opcode);
+ assert(ISD && "Invalid opcode");
+
+ std::pair<unsigned, MVT> LT = getTypeLegalizationCost(Ty);
+
+ if (!TLI->isOperationExpand(ISD, LT.second)) {
+ // The operation is legal. Assume it costs 1. Multiply
+ // by the type-legalization overhead.
+ return LT.first * 1;
+ }
+
+ // Else, assume that we need to scalarize this op.
+ if (Ty->isVectorTy()) {
+ unsigned Num = Ty->getVectorNumElements();
+ unsigned Cost = getArithmeticInstrCost(Opcode, Ty->getScalarType());
+ // return the cost of multiple scalar invocation plus the cost of inserting
+ // and extracting the values.
+ return getScalarizationOverhead(Ty, true, true) + Num * Cost;
+ }
+
+ // We don't know anything about this scalar instruction.
+ return 1;
+}
+
+unsigned VectorTargetTransformImpl::getBroadcastCost(Type *Tp) const {
+ return 1;
+}
+
+unsigned VectorTargetTransformImpl::getCastInstrCost(unsigned Opcode, Type *Dst,
+ Type *Src) const {
+ int ISD = InstructionOpcodeToISD(Opcode);
+ assert(ISD && "Invalid opcode");
+
+ std::pair<unsigned, MVT> SrcLT = getTypeLegalizationCost(Src);
+ std::pair<unsigned, MVT> DstLT = getTypeLegalizationCost(Dst);
+
+ // Handle scalar conversions.
+ if (!Src->isVectorTy() && !Dst->isVectorTy()) {
+
+ // Scalar bitcasts are usually free.
+ 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;
- // If we don't have any information about this instruction assume it costs 1.
- if (ISD == 0)
- return 1;
+ // Just check the op cost. If the operation is legal then assume it costs 1.
+ if (!TLI->isOperationExpand(ISD, DstLT.second))
+ return 1;
+
+ // Assume that illegal scalar instruction are expensive.
+ return 4;
+ }
+
+ // Check vector-to-vector casts.
+ if (Dst->isVectorTy() && Src->isVectorTy()) {
+
+ // If the cast is between same-sized registers, then the check is simple.
+ 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;
+
+ // Assume that sext is done using SHL and SRA.
+ if (Opcode == Instruction::SExt)
+ return 2;
+
+ // Just check the op cost. If the operation is legal then assume it costs
+ // 1 and multiply by the type-legalization overhead.
+ if (!TLI->isOperationExpand(ISD, DstLT.second))
+ return SrcLT.first * 1;
+ }
+
+ // If we are converting vectors and the operation is illegal, or
+ // if the vectors are legalized to different types, estimate the
+ // scalarization costs.
+ unsigned Num = Dst->getVectorNumElements();
+ unsigned Cost = getCastInstrCost(Opcode, Dst->getScalarType(),
+ Src->getScalarType());
+
+ // Return the cost of multiple scalar invocation plus the cost of
+ // inserting and extracting the values.
+ return getScalarizationOverhead(Dst, true, true) + Num * Cost;
+ }
+
+ // We already handled vector-to-vector and scalar-to-scalar conversions. This
+ // is where we handle bitcast between vectors and scalars. We need to assume
+ // that the conversion is scalarized in one way or another.
+ if (Opcode == Instruction::BitCast)
+ // Illegal bitcasts are done by storing and loading from a stack slot.
+ return (Src->isVectorTy()? getScalarizationOverhead(Src, false, true):0) +
+ (Dst->isVectorTy()? getScalarizationOverhead(Dst, true, false):0);
+
+ llvm_unreachable("Unhandled cast");
+ }
+
+unsigned VectorTargetTransformImpl::getCFInstrCost(unsigned Opcode) const {
+ // Branches are assumed to be predicted.
+ return 0;
+}
+
+unsigned VectorTargetTransformImpl::getCmpSelInstrCost(unsigned Opcode,
+ Type *ValTy,
+ Type *CondTy) const {
+ int ISD = InstructionOpcodeToISD(Opcode);
+ assert(ISD && "Invalid opcode");
// Selects on vectors are actually vector selects.
if (ISD == ISD::SELECT) {
- assert(Ty2 && "Ty2 must hold the condition type");
- if (Ty2->isVectorTy())
- ISD = ISD::VSELECT;
+ assert(CondTy && "CondTy must exist");
+ if (CondTy->isVectorTy())
+ ISD = ISD::VSELECT;
}
- assert(Ty1 && "We need to have at least one type");
+ std::pair<unsigned, MVT> LT = getTypeLegalizationCost(ValTy);
- // From this stage we look at the legalized type.
- std::pair<unsigned, EVT> LT =
- getTypeLegalizationCost(Ty1->getContext(), TLI->getValueType(Ty1));
-
- if (TLI->isOperationLegalOrCustom(ISD, LT.second)) {
+ if (!TLI->isOperationExpand(ISD, LT.second)) {
// The operation is legal. Assume it costs 1. Multiply
// by the type-legalization overhead.
return LT.first * 1;
}
- unsigned NumElem =
- (LT.second.isVector() ? LT.second.getVectorNumElements() : 1);
+ // Otherwise, assume that the cast is scalarized.
+ if (ValTy->isVectorTy()) {
+ unsigned Num = ValTy->getVectorNumElements();
+ if (CondTy)
+ CondTy = CondTy->getScalarType();
+ unsigned Cost = getCmpSelInstrCost(Opcode, ValTy->getScalarType(),
+ CondTy);
- // We will probably scalarize this instruction. Assume that the cost is the
- // number of the vector elements.
- return LT.first * NumElem * 1;
+ // Return the cost of multiple scalar invocation plus the cost of inserting
+ // and extracting the values.
+ return getScalarizationOverhead(ValTy, true, false) + Num * Cost;
+ }
+
+ // Unknown scalar opcode.
+ return 1;
}
-unsigned
-VectorTargetTransformImpl::getBroadcastCost(Type *Tp) const {
+unsigned VectorTargetTransformImpl::getVectorInstrCost(unsigned Opcode,
+ Type *Val,
+ unsigned Index) const {
return 1;
}
VectorTargetTransformImpl::getMemoryOpCost(unsigned Opcode, Type *Src,
unsigned Alignment,
unsigned AddressSpace) const {
- // From this stage we look at the legalized type.
- std::pair<unsigned, EVT> LT =
- getTypeLegalizationCost(Src->getContext(), TLI->getValueType(Src));
+ assert(!Src->isVoidTy() && "Invalid type");
+ std::pair<unsigned, MVT> LT = getTypeLegalizationCost(Src);
+
// Assume that all loads of legal types cost 1.
return LT.first;
}
+unsigned
+VectorTargetTransformImpl::getIntrinsicInstrCost(Intrinsic::ID, 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);
+ ScalarCalls = std::max(ScalarCalls, RetTy->getVectorNumElements());
+ }
+ }
+ return ScalarCalls + ScalarizationCost;
+}
+
unsigned
VectorTargetTransformImpl::getNumberOfParts(Type *Tp) const {
- std::pair<unsigned, EVT> LT =
- getTypeLegalizationCost(Tp->getContext(), TLI->getValueType(Tp));
+ std::pair<unsigned, MVT> LT = getTypeLegalizationCost(Tp);
return LT.first;
}
-
projects / oota-llvm.git / blobdiff