[LLVM] Adjust argument passing for APInt. NFC.
#81814
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
llvmbot
added
backend:ARM
backend:AArch64
backend:AMDGPU
backend:X86
mc
|
@llvm/pr-subscribers-llvm-globalisel @llvm/pr-subscribers-backend-arm Author: Yingwei Zheng (dtcxzyw) ChangesThis patch adjusts argument passing for
Compile-time impact: https://llvm-compile-time-tracker.com/compare.php?from=f82e0809ba12170e2f648f8a1ac01e78ef06c958&to=1b5e1cc27f66620860c9537e0bf7a537769494b1&stat=instructions:u Patch is 30.50 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/81814.diff 26 Files Affected:
diff --git a/llvm/include/llvm/Analysis/InlineCost.h b/llvm/include/llvm/Analysis/InlineCost.h
index 3a760e0a85cecf..c5978ce54fc18b 100644
--- a/llvm/include/llvm/Analysis/InlineCost.h
+++ b/llvm/include/llvm/Analysis/InlineCost.h
@@ -65,7 +65,8 @@ const char MaxInlineStackSizeAttributeName[] = "inline-max-stacksize";
// The cost-benefit pair computed by cost-benefit analysis.
class CostBenefitPair {
public:
- CostBenefitPair(APInt Cost, APInt Benefit) : Cost(Cost), Benefit(Benefit) {}
+ CostBenefitPair(APInt Cost, APInt Benefit)
+ : Cost(std::move(Cost)), Benefit(std::move(Benefit)) {}
const APInt &getCost() const { return Cost; }
diff --git a/llvm/include/llvm/Analysis/MemoryBuiltins.h b/llvm/include/llvm/Analysis/MemoryBuiltins.h
index 37ce1518f00c08..bb282a1b73d320 100644
--- a/llvm/include/llvm/Analysis/MemoryBuiltins.h
+++ b/llvm/include/llvm/Analysis/MemoryBuiltins.h
@@ -196,7 +196,8 @@ template <typename T, class C> struct SizeOffsetType {
T Offset;
SizeOffsetType() = default;
- SizeOffsetType(T Size, T Offset) : Size(Size), Offset(Offset) {}
+ SizeOffsetType(T Size, T Offset)
+ : Size(std::move(Size)), Offset(std::move(Offset)) {}
bool knownSize() const { return C::known(Size); }
bool knownOffset() const { return C::known(Offset); }
@@ -215,9 +216,10 @@ template <typename T, class C> struct SizeOffsetType {
/// \p APInts.
struct SizeOffsetAPInt : public SizeOffsetType<APInt, SizeOffsetAPInt> {
SizeOffsetAPInt() = default;
- SizeOffsetAPInt(APInt Size, APInt Offset) : SizeOffsetType(Size, Offset) {}
+ SizeOffsetAPInt(APInt Size, APInt Offset)
+ : SizeOffsetType(std::move(Size), std::move(Offset)) {}
- static bool known(APInt V) { return V.getBitWidth() > 1; }
+ static bool known(const APInt &V) { return V.getBitWidth() > 1; }
};
/// Evaluate the size and offset of an object pointed to by a Value*
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h b/llvm/include/llvm/Analysis/TargetTransformInfo.h
index 58577a6b6eb5c0..46cdd69fcd48e4 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfo.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h
@@ -666,14 +666,15 @@ class TargetTransformInfo {
IntrinsicInst & II) const;
/// Can be used to implement target-specific instruction combining.
/// \see instCombineIntrinsic
- std::optional<Value *> simplifyDemandedUseBitsIntrinsic(
- InstCombiner & IC, IntrinsicInst & II, APInt DemandedMask,
- KnownBits & Known, bool &KnownBitsComputed) const;
+ std::optional<Value *>
+ simplifyDemandedUseBitsIntrinsic(InstCombiner &IC, IntrinsicInst &II,
+ const APInt &DemandedMask, KnownBits &Known,
+ bool &KnownBitsComputed) const;
/// Can be used to implement target-specific instruction combining.
/// \see instCombineIntrinsic
std::optional<Value *> simplifyDemandedVectorEltsIntrinsic(
- InstCombiner & IC, IntrinsicInst & II, APInt DemandedElts,
- APInt & UndefElts, APInt & UndefElts2, APInt & UndefElts3,
+ InstCombiner &IC, IntrinsicInst &II, const APInt &DemandedElts,
+ APInt &UndefElts, APInt &UndefElts2, APInt &UndefElts3,
std::function<void(Instruction *, unsigned, APInt, APInt &)>
SimplifyAndSetOp) const;
/// @}
@@ -1826,11 +1827,12 @@ class TargetTransformInfo::Concept {
getPreferredTailFoldingStyle(bool IVUpdateMayOverflow = true) = 0;
virtual std::optional<Instruction *> instCombineIntrinsic(
InstCombiner &IC, IntrinsicInst &II) = 0;
- virtual std::optional<Value *> simplifyDemandedUseBitsIntrinsic(
- InstCombiner &IC, IntrinsicInst &II, APInt DemandedMask,
- KnownBits & Known, bool &KnownBitsComputed) = 0;
+ virtual std::optional<Value *>
+ simplifyDemandedUseBitsIntrinsic(InstCombiner &IC, IntrinsicInst &II,
+ const APInt &DemandedMask, KnownBits &Known,
+ bool &KnownBitsComputed) = 0;
virtual std::optional<Value *> simplifyDemandedVectorEltsIntrinsic(
- InstCombiner &IC, IntrinsicInst &II, APInt DemandedElts,
+ InstCombiner &IC, IntrinsicInst &II, const APInt &DemandedElts,
APInt &UndefElts, APInt &UndefElts2, APInt &UndefElts3,
std::function<void(Instruction *, unsigned, APInt, APInt &)>
SimplifyAndSetOp) = 0;
@@ -2278,14 +2280,14 @@ class TargetTransformInfo::Model final : public TargetTransformInfo::Concept {
}
std::optional<Value *>
simplifyDemandedUseBitsIntrinsic(InstCombiner &IC, IntrinsicInst &II,
- APInt DemandedMask, KnownBits &Known,
+ const APInt &DemandedMask, KnownBits &Known,
bool &KnownBitsComputed) override {
return Impl.simplifyDemandedUseBitsIntrinsic(IC, II, DemandedMask, Known,
KnownBitsComputed);
}
std::optional<Value *> simplifyDemandedVectorEltsIntrinsic(
- InstCombiner &IC, IntrinsicInst &II, APInt DemandedElts, APInt &UndefElts,
- APInt &UndefElts2, APInt &UndefElts3,
+ InstCombiner &IC, IntrinsicInst &II, const APInt &DemandedElts,
+ APInt &UndefElts, APInt &UndefElts2, APInt &UndefElts3,
std::function<void(Instruction *, unsigned, APInt, APInt &)>
SimplifyAndSetOp) override {
return Impl.simplifyDemandedVectorEltsIntrinsic(
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
index 3d5db96e86b804..48b32afc63d805 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
@@ -194,14 +194,14 @@ class TargetTransformInfoImplBase {
std::optional<Value *>
simplifyDemandedUseBitsIntrinsic(InstCombiner &IC, IntrinsicInst &II,
- APInt DemandedMask, KnownBits &Known,
+ const APInt &DemandedMask, KnownBits &Known,
bool &KnownBitsComputed) const {
return std::nullopt;
}
std::optional<Value *> simplifyDemandedVectorEltsIntrinsic(
- InstCombiner &IC, IntrinsicInst &II, APInt DemandedElts, APInt &UndefElts,
- APInt &UndefElts2, APInt &UndefElts3,
+ InstCombiner &IC, IntrinsicInst &II, const APInt &DemandedElts,
+ APInt &UndefElts, APInt &UndefElts2, APInt &UndefElts3,
std::function<void(Instruction *, unsigned, APInt, APInt &)>
SimplifyAndSetOp) const {
return std::nullopt;
diff --git a/llvm/include/llvm/CodeGen/BasicTTIImpl.h b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
index bb17298daba03a..b5cb86c9f46f27 100644
--- a/llvm/include/llvm/CodeGen/BasicTTIImpl.h
+++ b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
@@ -675,15 +675,15 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
std::optional<Value *>
simplifyDemandedUseBitsIntrinsic(InstCombiner &IC, IntrinsicInst &II,
- APInt DemandedMask, KnownBits &Known,
+ const APInt &DemandedMask, KnownBits &Known,
bool &KnownBitsComputed) {
return BaseT::simplifyDemandedUseBitsIntrinsic(IC, II, DemandedMask, Known,
KnownBitsComputed);
}
std::optional<Value *> simplifyDemandedVectorEltsIntrinsic(
- InstCombiner &IC, IntrinsicInst &II, APInt DemandedElts, APInt &UndefElts,
- APInt &UndefElts2, APInt &UndefElts3,
+ InstCombiner &IC, IntrinsicInst &II, const APInt &DemandedElts,
+ APInt &UndefElts, APInt &UndefElts2, APInt &UndefElts3,
std::function<void(Instruction *, unsigned, APInt, APInt &)>
SimplifyAndSetOp) {
return BaseT::simplifyDemandedVectorEltsIntrinsic(
diff --git a/llvm/include/llvm/CodeGen/SelectionDAG.h b/llvm/include/llvm/CodeGen/SelectionDAG.h
index 7bb12d8f065c9d..8467e451d766d7 100644
--- a/llvm/include/llvm/CodeGen/SelectionDAG.h
+++ b/llvm/include/llvm/CodeGen/SelectionDAG.h
@@ -883,7 +883,7 @@ class SelectionDAG {
/// Returns a vector of type ResVT whose elements contain the linear sequence
/// <0, Step, Step * 2, Step * 3, ...>
- SDValue getStepVector(const SDLoc &DL, EVT ResVT, APInt StepVal);
+ SDValue getStepVector(const SDLoc &DL, EVT ResVT, const APInt &StepVal);
/// Returns a vector of type ResVT whose elements contain the linear sequence
/// <0, 1, 2, 3, ...>
diff --git a/llvm/include/llvm/IR/PatternMatch.h b/llvm/include/llvm/IR/PatternMatch.h
index fed552414298ad..785258a1be8625 100644
--- a/llvm/include/llvm/IR/PatternMatch.h
+++ b/llvm/include/llvm/IR/PatternMatch.h
@@ -843,9 +843,9 @@ struct bind_const_intval_ty {
/// Match a specified integer value or vector of all elements of that
/// value.
template <bool AllowUndefs> struct specific_intval {
- APInt Val;
+ const APInt &Val;
- specific_intval(APInt V) : Val(std::move(V)) {}
+ specific_intval(const APInt &V) : Val(V) {}
template <typename ITy> bool match(ITy *V) {
const auto *CI = dyn_cast<ConstantInt>(V);
@@ -857,22 +857,37 @@ template <bool AllowUndefs> struct specific_intval {
}
};
+template <bool AllowUndefs> struct specific_intval64 {
+ uint64_t Val;
+
+ specific_intval64(uint64_t V) : Val(V) {}
+
+ template <typename ITy> bool match(ITy *V) {
+ const auto *CI = dyn_cast<ConstantInt>(V);
+ if (!CI && V->getType()->isVectorTy())
+ if (const auto *C = dyn_cast<Constant>(V))
+ CI = dyn_cast_or_null<ConstantInt>(C->getSplatValue(AllowUndefs));
+
+ return CI && CI->getValue() == Val;
+ }
+};
+
/// Match a specific integer value or vector with all elements equal to
/// the value.
-inline specific_intval<false> m_SpecificInt(APInt V) {
- return specific_intval<false>(std::move(V));
+inline specific_intval<false> m_SpecificInt(const APInt &V) {
+ return specific_intval<false>(V);
}
-inline specific_intval<false> m_SpecificInt(uint64_t V) {
- return m_SpecificInt(APInt(64, V));
+inline specific_intval64<false> m_SpecificInt(uint64_t V) {
+ return specific_intval64<false>(V);
}
-inline specific_intval<true> m_SpecificIntAllowUndef(APInt V) {
- return specific_intval<true>(std::move(V));
+inline specific_intval<true> m_SpecificIntAllowUndef(const APInt &V) {
+ return specific_intval<true>(V);
}
-inline specific_intval<true> m_SpecificIntAllowUndef(uint64_t V) {
- return m_SpecificIntAllowUndef(APInt(64, V));
+inline specific_intval64<true> m_SpecificIntAllowUndef(uint64_t V) {
+ return specific_intval64<true>(V);
}
/// Match a ConstantInt and bind to its value. This does not match
diff --git a/llvm/include/llvm/MC/MCStreamer.h b/llvm/include/llvm/MC/MCStreamer.h
index 671511ab4b88e3..69867620e1bf8a 100644
--- a/llvm/include/llvm/MC/MCStreamer.h
+++ b/llvm/include/llvm/MC/MCStreamer.h
@@ -740,7 +740,7 @@ class MCStreamer {
/// Special case of EmitValue that avoids the client having
/// to pass in a MCExpr for constant integers.
virtual void emitIntValue(uint64_t Value, unsigned Size);
- virtual void emitIntValue(APInt Value);
+ virtual void emitIntValue(const APInt &Value);
/// Special case of EmitValue that avoids the client having to pass
/// in a MCExpr for constant integers & prints in Hex format for certain
diff --git a/llvm/include/llvm/Transforms/InstCombine/InstCombiner.h b/llvm/include/llvm/Transforms/InstCombine/InstCombiner.h
index 93090431cbb69f..d910b51a90a980 100644
--- a/llvm/include/llvm/Transforms/InstCombine/InstCombiner.h
+++ b/llvm/include/llvm/Transforms/InstCombine/InstCombiner.h
@@ -348,12 +348,11 @@ class LLVM_LIBRARY_VISIBILITY InstCombiner {
// Call target specific combiners
std::optional<Instruction *> targetInstCombineIntrinsic(IntrinsicInst &II);
- std::optional<Value *>
- targetSimplifyDemandedUseBitsIntrinsic(IntrinsicInst &II, APInt DemandedMask,
- KnownBits &Known,
- bool &KnownBitsComputed);
+ std::optional<Value *> targetSimplifyDemandedUseBitsIntrinsic(
+ IntrinsicInst &II, const APInt &DemandedMask, KnownBits &Known,
+ bool &KnownBitsComputed);
std::optional<Value *> targetSimplifyDemandedVectorEltsIntrinsic(
- IntrinsicInst &II, APInt DemandedElts, APInt &UndefElts,
+ IntrinsicInst &II, const APInt &DemandedElts, APInt &UndefElts,
APInt &UndefElts2, APInt &UndefElts3,
std::function<void(Instruction *, unsigned, APInt, APInt &)>
SimplifyAndSetOp);
diff --git a/llvm/lib/Analysis/ConstantFolding.cpp b/llvm/lib/Analysis/ConstantFolding.cpp
index 90da3390eab324..c27b63f57a2041 100644
--- a/llvm/lib/Analysis/ConstantFolding.cpp
+++ b/llvm/lib/Analysis/ConstantFolding.cpp
@@ -751,7 +751,7 @@ Constant *llvm::ConstantFoldLoadFromConstPtr(Constant *C, Type *Ty,
Constant *llvm::ConstantFoldLoadFromConstPtr(Constant *C, Type *Ty,
const DataLayout &DL) {
APInt Offset(DL.getIndexTypeSizeInBits(C->getType()), 0);
- return ConstantFoldLoadFromConstPtr(C, Ty, Offset, DL);
+ return ConstantFoldLoadFromConstPtr(C, Ty, std::move(Offset), DL);
}
Constant *llvm::ConstantFoldLoadFromUniformValue(Constant *C, Type *Ty) {
diff --git a/llvm/lib/Analysis/InstructionSimplify.cpp b/llvm/lib/Analysis/InstructionSimplify.cpp
index 08050becd2df88..8f737b85fd2ebc 100644
--- a/llvm/lib/Analysis/InstructionSimplify.cpp
+++ b/llvm/lib/Analysis/InstructionSimplify.cpp
@@ -6105,7 +6105,8 @@ static Value *simplifyRelativeLoad(Constant *Ptr, Constant *Offset,
if (OffsetInt.srem(4) != 0)
return nullptr;
- Constant *Loaded = ConstantFoldLoadFromConstPtr(Ptr, Int32Ty, OffsetInt, DL);
+ Constant *Loaded =
+ ConstantFoldLoadFromConstPtr(Ptr, Int32Ty, std::move(OffsetInt), DL);
if (!Loaded)
return nullptr;
@@ -6973,7 +6974,8 @@ Value *llvm::simplifyLoadInst(LoadInst *LI, Value *PtrOp,
if (PtrOp == GV) {
// Index size may have changed due to address space casts.
Offset = Offset.sextOrTrunc(Q.DL.getIndexTypeSizeInBits(PtrOp->getType()));
- return ConstantFoldLoadFromConstPtr(GV, LI->getType(), Offset, Q.DL);
+ return ConstantFoldLoadFromConstPtr(GV, LI->getType(), std::move(Offset),
+ Q.DL);
}
return nullptr;
diff --git a/llvm/lib/Analysis/TargetTransformInfo.cpp b/llvm/lib/Analysis/TargetTransformInfo.cpp
index 1f11f0d7dd620e..ff4c6436f1f5b7 100644
--- a/llvm/lib/Analysis/TargetTransformInfo.cpp
+++ b/llvm/lib/Analysis/TargetTransformInfo.cpp
@@ -364,15 +364,15 @@ TargetTransformInfo::instCombineIntrinsic(InstCombiner &IC,
}
std::optional<Value *> TargetTransformInfo::simplifyDemandedUseBitsIntrinsic(
- InstCombiner &IC, IntrinsicInst &II, APInt DemandedMask, KnownBits &Known,
- bool &KnownBitsComputed) const {
+ InstCombiner &IC, IntrinsicInst &II, const APInt &DemandedMask,
+ KnownBits &Known, bool &KnownBitsComputed) const {
return TTIImpl->simplifyDemandedUseBitsIntrinsic(IC, II, DemandedMask, Known,
KnownBitsComputed);
}
std::optional<Value *> TargetTransformInfo::simplifyDemandedVectorEltsIntrinsic(
- InstCombiner &IC, IntrinsicInst &II, APInt DemandedElts, APInt &UndefElts,
- APInt &UndefElts2, APInt &UndefElts3,
+ InstCombiner &IC, IntrinsicInst &II, const APInt &DemandedElts,
+ APInt &UndefElts, APInt &UndefElts2, APInt &UndefElts3,
std::function<void(Instruction *, unsigned, APInt, APInt &)>
SimplifyAndSetOp) const {
return TTIImpl->simplifyDemandedVectorEltsIntrinsic(
diff --git a/llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp b/llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp
index e39fdae1ccbedb..a10d98a81c9c57 100644
--- a/llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp
+++ b/llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp
@@ -7697,7 +7697,7 @@ LegalizerHelper::LegalizeResult LegalizerHelper::lowerBswap(MachineInstr &MI) {
//{ (Src & Mask) >> N } | { (Src << N) & Mask }
static MachineInstrBuilder SwapN(unsigned N, DstOp Dst, MachineIRBuilder &B,
- MachineInstrBuilder Src, APInt Mask) {
+ MachineInstrBuilder Src, const APInt &Mask) {
const LLT Ty = Dst.getLLTTy(*B.getMRI());
MachineInstrBuilder C_N = B.buildConstant(Ty, N);
MachineInstrBuilder MaskLoNTo0 = B.buildConstant(Ty, Mask);
diff --git a/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp b/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
index 421bb516ad242f..3b3281c4df245e 100644
--- a/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
@@ -2029,7 +2029,8 @@ SDValue SelectionDAG::getStepVector(const SDLoc &DL, EVT ResVT) {
return getStepVector(DL, ResVT, One);
}
-SDValue SelectionDAG::getStepVector(const SDLoc &DL, EVT ResVT, APInt StepVal) {
+SDValue SelectionDAG::getStepVector(const SDLoc &DL, EVT ResVT,
+ const APInt &StepVal) {
assert(ResVT.getScalarSizeInBits() == StepVal.getBitWidth());
if (ResVT.isScalableVector())
return getNode(
diff --git a/llvm/lib/IR/LLVMContextImpl.h b/llvm/lib/IR/LLVMContextImpl.h
index 6a20291344989d..1185484b5e8ada 100644
--- a/llvm/lib/IR/LLVMContextImpl.h
+++ b/llvm/lib/IR/LLVMContextImpl.h
@@ -440,7 +440,7 @@ template <> struct MDNodeKeyImpl<DIEnumerator> {
bool IsUnsigned;
MDNodeKeyImpl(APInt Value, bool IsUnsigned, MDString *Name)
- : Value(Value), Name(Name), IsUnsigned(IsUnsigned) {}
+ : Value(std::move(Value)), Name(Name), IsUnsigned(IsUnsigned) {}
MDNodeKeyImpl(int64_t Value, bool IsUnsigned, MDString *Name)
: Value(APInt(64, Value, !IsUnsigned)), Name(Name),
IsUnsigned(IsUnsigned) {}
diff --git a/llvm/lib/MC/MCStreamer.cpp b/llvm/lib/MC/MCStreamer.cpp
index d0395770ae8b05..176d55aa890bed 100644
--- a/llvm/lib/MC/MCStreamer.cpp
+++ b/llvm/lib/MC/MCStreamer.cpp
@@ -141,7 +141,7 @@ void MCStreamer::emitIntValue(uint64_t Value, unsigned Size) {
unsigned Index = IsLittleEndian ? 0 : 8 - Size;
emitBytes(StringRef(reinterpret_cast<char *>(&Swapped) + Index, Size));
}
-void MCStreamer::emitIntValue(APInt Value) {
+void MCStreamer::emitIntValue(const APInt &Value) {
if (Value.getNumWords() == 1) {
emitIntValue(Value.getLimitedValue(), Value.getBitWidth() / 8);
return;
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
index af0b6873d170dd..25ace35fa7d701 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -2037,7 +2037,7 @@ AArch64TTIImpl::instCombineIntrinsic(InstCombiner &IC,
}
std::optional<Value *> AArch64TTIImpl::simplifyDemandedVectorEltsIntrinsic(
- InstCombiner &IC, IntrinsicInst &II, APInt OrigDemandedElts,
+ InstCombiner &IC, IntrinsicInst &II, const APInt &OrigDemandedElts,
APInt &UndefElts, APInt &UndefElts2, APInt &UndefElts3,
std::function<void(Instruction *, unsigned, APInt, APInt &)>
SimplifyAndSetOp) const {
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
index de39dea2be43e1..1a50f299091e3b 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
@@ -125,8 +125,8 @@ class AArch64TTIImpl : public BasicTTIImplBase<AArch64TTIImpl> {
IntrinsicInst &II) const;
std::optional<Value *> simplifyDemandedVectorEltsIntrinsic(
- InstCombiner &IC, IntrinsicInst &II, APInt DemandedElts, APInt &UndefElts,
- APInt &UndefElts2, APInt &UndefElts3,
+ InstCombiner &IC, IntrinsicInst &II, const APInt &DemandedElts,
+ APInt &UndefElts, APInt &UndefElts2, APInt &UndefElts3,
std::function<void(Instruction *, unsigned, APInt, APInt &)>
SimplifyAndSetOp) const;
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp b/llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp
index fb829fab0a2c19..0d1f65d66f0157 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp
@@ -1368,8 +1...
[truncated]
|
nikic
reviewed
Feb 15, 2024
llvm/include/llvm/IR/PatternMatch.h
Outdated
| @@ -843,9 +843,9 @@ struct bind_const_intval_ty { | |||
| /// Match a specified integer value or vector of all elements of that | |||
| /// value. | |||
| template <bool AllowUndefs> struct specific_intval { | |||
| APInt Val; | |||
| const APInt &Val; | |||
There was a problem hiding this comment.
Hm, is this safe for all usages? Might the APInt be temporary?
There was a problem hiding this comment.
It only needs to be alive through the m_SpecificInt call where the temporary would be created in the match expression
There was a problem hiding this comment.
Doesn't it have to be alive for the whole match() call, not just the m_SpecificInt() call? Does temporary promotion take care of that?
RKSimon
reviewed
Feb 29, 2024
There was a problem hiding this comment.
No objections, but splitting this into separate focused patches might be cleaner:
- simplifyDemandedVectorEltsIntrinsic refactor
- PatternMatch change [PatternMatch] Refactor
m_SpecificIntto avoid constructing APInt. NFC. #86259 - remaining const APInt& arguments [LLVM] Pass APInt by const reference. NFC. #86278
- use of std::move in constructors [LLVM] Use
std::movefor APInt. NFC. #86257
|
@dtcxzyw would you prefer to rebase this now that most split off changes are in or close and move remaining diffs to a new PR? |
|
Compile-time impact: https://llvm-compile-time-tracker.com/compare.php?from=d76a1233f7d7923d056a53cfa6f89735e9cda86e&to=69dbcd04b04d10a647e148ee68816142034bb6fb&stat=instructions:u The result looks not promising. Should I close this PR? @RKSimon |
|
From a refactor / cleanup the simplifyDemandedVectorEltsIntrinsic change still seems like the right direction - but those stats suggest its not going to actually affect compile times. I imagine we've barely got any vector masks > 64 elements |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This patch adjusts argument passing for
APIntto improve the compile-time.std::move.Compile-time impact: https://llvm-compile-time-tracker.com/compare.php?from=f82e0809ba12170e2f648f8a1ac01e78ef06c958&to=1b5e1cc27f66620860c9537e0bf7a537769494b1&stat=instructions:u