//===- LoopVectorize.cpp - A Loop Vectorizer ------------------------------===// // // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. // See https://llvm.org/LICENSE.txt for license information. // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception // //===----------------------------------------------------------------------===// // // This is the LLVM loop vectorizer. This pass modifies 'vectorizable' loops // and generates target-independent LLVM-IR. // The vectorizer uses the TargetTransformInfo analysis to estimate the costs // of instructions in order to estimate the profitability of vectorization. // // The loop vectorizer combines consecutive loop iterations into a single // 'wide' iteration. After this transformation the index is incremented // by the SIMD vector width, and not by one. // // This pass has three parts: // 1. The main loop pass that drives the different parts. // 2. LoopVectorizationLegality - A unit that checks for the legality // of the vectorization. // 3. InnerLoopVectorizer - A unit that performs the actual // widening of instructions. // 4. LoopVectorizationCostModel - A unit that checks for the profitability // of vectorization. It decides on the optimal vector width, which // can be one, if vectorization is not profitable. // // There is a development effort going on to migrate loop vectorizer to the // VPlan infrastructure and to introduce outer loop vectorization support (see // docs/VectorizationPlan.rst and // http://lists.llvm.org/pipermail/llvm-dev/2017-December/119523.html). For this // purpose, we temporarily introduced the VPlan-native vectorization path: an // alternative vectorization path that is natively implemented on top of the // VPlan infrastructure. See EnableVPlanNativePath for enabling. // //===----------------------------------------------------------------------===// // // The reduction-variable vectorization is based on the paper: // D. Nuzman and R. Henderson. Multi-platform Auto-vectorization. // // Variable uniformity checks are inspired by: // Karrenberg, R. and Hack, S. Whole Function Vectorization. // // The interleaved access vectorization is based on the paper: // Dorit Nuzman, Ira Rosen and Ayal Zaks. Auto-Vectorization of Interleaved // Data for SIMD // // Other ideas/concepts are from: // A. Zaks and D. Nuzman. Autovectorization in GCC-two years later. // // S. Maleki, Y. Gao, M. Garzaran, T. Wong and D. Padua. An Evaluation of // Vectorizing Compilers. // //===----------------------------------------------------------------------===// #include "llvm/Transforms/Vectorize/LoopVectorize.h" #include "LoopVectorizationPlanner.h" #include "VPRecipeBuilder.h" #include "VPlan.h" #include "VPlanAnalysis.h" #include "VPlanHCFGBuilder.h" #include "VPlanTransforms.h" #include "llvm/ADT/APInt.h" #include "llvm/ADT/ArrayRef.h" #include "llvm/ADT/DenseMap.h" #include "llvm/ADT/DenseMapInfo.h" #include "llvm/ADT/Hashing.h" #include "llvm/ADT/MapVector.h" #include "llvm/ADT/STLExtras.h" #include "llvm/ADT/SmallPtrSet.h" #include "llvm/ADT/SmallSet.h" #include "llvm/ADT/SmallVector.h" #include "llvm/ADT/Statistic.h" #include "llvm/ADT/StringRef.h" #include "llvm/ADT/Twine.h" #include "llvm/ADT/iterator_range.h" #include "llvm/Analysis/AssumptionCache.h" #include "llvm/Analysis/BasicAliasAnalysis.h" #include "llvm/Analysis/BlockFrequencyInfo.h" #include "llvm/Analysis/CFG.h" #include "llvm/Analysis/CodeMetrics.h" #include "llvm/Analysis/DemandedBits.h" #include "llvm/Analysis/GlobalsModRef.h" #include "llvm/Analysis/LoopAccessAnalysis.h" #include "llvm/Analysis/LoopAnalysisManager.h" #include "llvm/Analysis/LoopInfo.h" #include "llvm/Analysis/LoopIterator.h" #include "llvm/Analysis/OptimizationRemarkEmitter.h" #include "llvm/Analysis/ProfileSummaryInfo.h" #include "llvm/Analysis/ScalarEvolution.h" #include "llvm/Analysis/ScalarEvolutionExpressions.h" #include "llvm/Analysis/TargetLibraryInfo.h" #include "llvm/Analysis/TargetTransformInfo.h" #include "llvm/Analysis/ValueTracking.h" #include "llvm/Analysis/VectorUtils.h" #include "llvm/IR/Attributes.h" #include "llvm/IR/BasicBlock.h" #include "llvm/IR/CFG.h" #include "llvm/IR/Constant.h" #include "llvm/IR/Constants.h" #include "llvm/IR/DataLayout.h" #include "llvm/IR/DebugInfo.h" #include "llvm/IR/DebugInfoMetadata.h" #include "llvm/IR/DebugLoc.h" #include "llvm/IR/DerivedTypes.h" #include "llvm/IR/DiagnosticInfo.h" #include "llvm/IR/Dominators.h" #include "llvm/IR/Function.h" #include "llvm/IR/IRBuilder.h" #include "llvm/IR/InstrTypes.h" #include "llvm/IR/Instruction.h" #include "llvm/IR/Instructions.h" #include "llvm/IR/IntrinsicInst.h" #include "llvm/IR/Intrinsics.h" #include "llvm/IR/MDBuilder.h" #include "llvm/IR/Metadata.h" #include "llvm/IR/Module.h" #include "llvm/IR/Operator.h" #include "llvm/IR/PatternMatch.h" #include "llvm/IR/ProfDataUtils.h" #include "llvm/IR/Type.h" #include "llvm/IR/Use.h" #include "llvm/IR/User.h" #include "llvm/IR/Value.h" #include "llvm/IR/ValueHandle.h" #include "llvm/IR/Verifier.h" #include "llvm/Support/Casting.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/Compiler.h" #include "llvm/Support/Debug.h" #include "llvm/Support/ErrorHandling.h" #include "llvm/Support/InstructionCost.h" #include "llvm/Support/MathExtras.h" #include "llvm/Support/raw_ostream.h" #include "llvm/Transforms/Utils/BasicBlockUtils.h" #include "llvm/Transforms/Utils/InjectTLIMappings.h" #include "llvm/Transforms/Utils/LoopSimplify.h" #include "llvm/Transforms/Utils/LoopUtils.h" #include "llvm/Transforms/Utils/LoopVersioning.h" #include "llvm/Transforms/Utils/ScalarEvolutionExpander.h" #include "llvm/Transforms/Utils/SizeOpts.h" #include "llvm/Transforms/Vectorize/LoopVectorizationLegality.h" #include #include #include #include #include #include #include #include #include #include #include #include using namespace llvm; #define LV_NAME "loop-vectorize" #define DEBUG_TYPE LV_NAME #ifndef NDEBUG const char VerboseDebug[] = DEBUG_TYPE "-verbose"; #endif /// @{ /// Metadata attribute names const char LLVMLoopVectorizeFollowupAll[] = "llvm.loop.vectorize.followup_all"; const char LLVMLoopVectorizeFollowupVectorized[] = "llvm.loop.vectorize.followup_vectorized"; const char LLVMLoopVectorizeFollowupEpilogue[] = "llvm.loop.vectorize.followup_epilogue"; /// @} STATISTIC(LoopsVectorized, "Number of loops vectorized"); STATISTIC(LoopsAnalyzed, "Number of loops analyzed for vectorization"); STATISTIC(LoopsEpilogueVectorized, "Number of epilogues vectorized"); static cl::opt EnableEpilogueVectorization( "enable-epilogue-vectorization", cl::init(true), cl::Hidden, cl::desc("Enable vectorization of epilogue loops.")); static cl::opt EpilogueVectorizationForceVF( "epilogue-vectorization-force-VF", cl::init(1), cl::Hidden, cl::desc("When epilogue vectorization is enabled, and a value greater than " "1 is specified, forces the given VF for all applicable epilogue " "loops.")); static cl::opt EpilogueVectorizationMinVF( "epilogue-vectorization-minimum-VF", cl::init(16), cl::Hidden, cl::desc("Only loops with vectorization factor equal to or larger than " "the specified value are considered for epilogue vectorization.")); /// Loops with a known constant trip count below this number are vectorized only /// if no scalar iteration overheads are incurred. static cl::opt TinyTripCountVectorThreshold( "vectorizer-min-trip-count", cl::init(16), cl::Hidden, cl::desc("Loops with a constant trip count that is smaller than this " "value are vectorized only if no scalar iteration overheads " "are incurred.")); static cl::opt VectorizeMemoryCheckThreshold( "vectorize-memory-check-threshold", cl::init(128), cl::Hidden, cl::desc("The maximum allowed number of runtime memory checks")); // Option prefer-predicate-over-epilogue indicates that an epilogue is undesired, // that predication is preferred, and this lists all options. I.e., the // vectorizer will try to fold the tail-loop (epilogue) into the vector body // and predicate the instructions accordingly. If tail-folding fails, there are // different fallback strategies depending on these values: namespace PreferPredicateTy { enum Option { ScalarEpilogue = 0, PredicateElseScalarEpilogue, PredicateOrDontVectorize }; } // namespace PreferPredicateTy static cl::opt PreferPredicateOverEpilogue( "prefer-predicate-over-epilogue", cl::init(PreferPredicateTy::ScalarEpilogue), cl::Hidden, cl::desc("Tail-folding and predication preferences over creating a scalar " "epilogue loop."), cl::values(clEnumValN(PreferPredicateTy::ScalarEpilogue, "scalar-epilogue", "Don't tail-predicate loops, create scalar epilogue"), clEnumValN(PreferPredicateTy::PredicateElseScalarEpilogue, "predicate-else-scalar-epilogue", "prefer tail-folding, create scalar epilogue if tail " "folding fails."), clEnumValN(PreferPredicateTy::PredicateOrDontVectorize, "predicate-dont-vectorize", "prefers tail-folding, don't attempt vectorization if " "tail-folding fails."))); static cl::opt ForceTailFoldingStyle( "force-tail-folding-style", cl::desc("Force the tail folding style"), cl::init(TailFoldingStyle::None), cl::values( clEnumValN(TailFoldingStyle::None, "none", "Disable tail folding"), clEnumValN( TailFoldingStyle::Data, "data", "Create lane mask for data only, using active.lane.mask intrinsic"), clEnumValN(TailFoldingStyle::DataWithoutLaneMask, "data-without-lane-mask", "Create lane mask with compare/stepvector"), clEnumValN(TailFoldingStyle::DataAndControlFlow, "data-and-control", "Create lane mask using active.lane.mask intrinsic, and use " "it for both data and control flow"), clEnumValN( TailFoldingStyle::DataAndControlFlowWithoutRuntimeCheck, "data-and-control-without-rt-check", "Similar to data-and-control, but remove the runtime check"))); static cl::opt MaximizeBandwidth( "vectorizer-maximize-bandwidth", cl::init(false), cl::Hidden, cl::desc("Maximize bandwidth when selecting vectorization factor which " "will be determined by the smallest type in loop.")); static cl::opt EnableInterleavedMemAccesses( "enable-interleaved-mem-accesses", cl::init(false), cl::Hidden, cl::desc("Enable vectorization on interleaved memory accesses in a loop")); /// An interleave-group may need masking if it resides in a block that needs /// predication, or in order to mask away gaps. static cl::opt EnableMaskedInterleavedMemAccesses( "enable-masked-interleaved-mem-accesses", cl::init(false), cl::Hidden, cl::desc("Enable vectorization on masked interleaved memory accesses in a loop")); static cl::opt TinyTripCountInterleaveThreshold( "tiny-trip-count-interleave-threshold", cl::init(128), cl::Hidden, cl::desc("We don't interleave loops with a estimated constant trip count " "below this number")); static cl::opt ForceTargetNumScalarRegs( "force-target-num-scalar-regs", cl::init(0), cl::Hidden, cl::desc("A flag that overrides the target's number of scalar registers.")); static cl::opt ForceTargetNumVectorRegs( "force-target-num-vector-regs", cl::init(0), cl::Hidden, cl::desc("A flag that overrides the target's number of vector registers.")); static cl::opt ForceTargetMaxScalarInterleaveFactor( "force-target-max-scalar-interleave", cl::init(0), cl::Hidden, cl::desc("A flag that overrides the target's max interleave factor for " "scalar loops.")); static cl::opt ForceTargetMaxVectorInterleaveFactor( "force-target-max-vector-interleave", cl::init(0), cl::Hidden, cl::desc("A flag that overrides the target's max interleave factor for " "vectorized loops.")); static cl::opt ForceTargetInstructionCost( "force-target-instruction-cost", cl::init(0), cl::Hidden, cl::desc("A flag that overrides the target's expected cost for " "an instruction to a single constant value. Mostly " "useful for getting consistent testing.")); static cl::opt ForceTargetSupportsScalableVectors( "force-target-supports-scalable-vectors", cl::init(false), cl::Hidden, cl::desc( "Pretend that scalable vectors are supported, even if the target does " "not support them. This flag should only be used for testing.")); static cl::opt SmallLoopCost( "small-loop-cost", cl::init(20), cl::Hidden, cl::desc( "The cost of a loop that is considered 'small' by the interleaver.")); static cl::opt LoopVectorizeWithBlockFrequency( "loop-vectorize-with-block-frequency", cl::init(true), cl::Hidden, cl::desc("Enable the use of the block frequency analysis to access PGO " "heuristics minimizing code growth in cold regions and being more " "aggressive in hot regions.")); // Runtime interleave loops for load/store throughput. static cl::opt EnableLoadStoreRuntimeInterleave( "enable-loadstore-runtime-interleave", cl::init(true), cl::Hidden, cl::desc( "Enable runtime interleaving until load/store ports are saturated")); /// Interleave small loops with scalar reductions. static cl::opt InterleaveSmallLoopScalarReduction( "interleave-small-loop-scalar-reduction", cl::init(false), cl::Hidden, cl::desc("Enable interleaving for loops with small iteration counts that " "contain scalar reductions to expose ILP.")); /// The number of stores in a loop that are allowed to need predication. static cl::opt NumberOfStoresToPredicate( "vectorize-num-stores-pred", cl::init(1), cl::Hidden, cl::desc("Max number of stores to be predicated behind an if.")); static cl::opt EnableIndVarRegisterHeur( "enable-ind-var-reg-heur", cl::init(true), cl::Hidden, cl::desc("Count the induction variable only once when interleaving")); static cl::opt EnableCondStoresVectorization( "enable-cond-stores-vec", cl::init(true), cl::Hidden, cl::desc("Enable if predication of stores during vectorization.")); static cl::opt MaxNestedScalarReductionIC( "max-nested-scalar-reduction-interleave", cl::init(2), cl::Hidden, cl::desc("The maximum interleave count to use when interleaving a scalar " "reduction in a nested loop.")); static cl::opt PreferInLoopReductions("prefer-inloop-reductions", cl::init(false), cl::Hidden, cl::desc("Prefer in-loop vector reductions, " "overriding the targets preference.")); static cl::opt ForceOrderedReductions( "force-ordered-reductions", cl::init(false), cl::Hidden, cl::desc("Enable the vectorisation of loops with in-order (strict) " "FP reductions")); static cl::opt PreferPredicatedReductionSelect( "prefer-predicated-reduction-select", cl::init(false), cl::Hidden, cl::desc( "Prefer predicating a reduction operation over an after loop select.")); namespace llvm { cl::opt EnableVPlanNativePath( "enable-vplan-native-path", cl::Hidden, cl::desc("Enable VPlan-native vectorization path with " "support for outer loop vectorization.")); } // This flag enables the stress testing of the VPlan H-CFG construction in the // VPlan-native vectorization path. It must be used in conjuction with // -enable-vplan-native-path. -vplan-verify-hcfg can also be used to enable the // verification of the H-CFGs built. static cl::opt VPlanBuildStressTest( "vplan-build-stress-test", cl::init(false), cl::Hidden, cl::desc( "Build VPlan for every supported loop nest in the function and bail " "out right after the build (stress test the VPlan H-CFG construction " "in the VPlan-native vectorization path).")); cl::opt llvm::EnableLoopInterleaving( "interleave-loops", cl::init(true), cl::Hidden, cl::desc("Enable loop interleaving in Loop vectorization passes")); cl::opt llvm::EnableLoopVectorization( "vectorize-loops", cl::init(true), cl::Hidden, cl::desc("Run the Loop vectorization passes")); static cl::opt PrintVPlansInDotFormat( "vplan-print-in-dot-format", cl::Hidden, cl::desc("Use dot format instead of plain text when dumping VPlans")); static cl::opt ForceSafeDivisor( "force-widen-divrem-via-safe-divisor", cl::Hidden, cl::desc( "Override cost based safe divisor widening for div/rem instructions")); static cl::opt UseWiderVFIfCallVariantsPresent( "vectorizer-maximize-bandwidth-for-vector-calls", cl::init(true), cl::Hidden, cl::desc("Try wider VFs if they enable the use of vector variants")); // Likelyhood of bypassing the vectorized loop because assumptions about SCEV // variables not overflowing do not hold. See `emitSCEVChecks`. static constexpr uint32_t SCEVCheckBypassWeights[] = {1, 127}; // Likelyhood of bypassing the vectorized loop because pointers overlap. See // `emitMemRuntimeChecks`. static constexpr uint32_t MemCheckBypassWeights[] = {1, 127}; // Likelyhood of bypassing the vectorized loop because there are zero trips left // after prolog. See `emitIterationCountCheck`. static constexpr uint32_t MinItersBypassWeights[] = {1, 127}; /// A helper function that returns true if the given type is irregular. The /// type is irregular if its allocated size doesn't equal the store size of an /// element of the corresponding vector type. static bool hasIrregularType(Type *Ty, const DataLayout &DL) { // Determine if an array of N elements of type Ty is "bitcast compatible" // with a vector. // This is only true if there is no padding between the array elements. return DL.getTypeAllocSizeInBits(Ty) != DL.getTypeSizeInBits(Ty); } /// A helper function that returns the reciprocal of the block probability of /// predicated blocks. If we return X, we are assuming the predicated block /// will execute once for every X iterations of the loop header. /// /// TODO: We should use actual block probability here, if available. Currently, /// we always assume predicated blocks have a 50% chance of executing. static unsigned getReciprocalPredBlockProb() { return 2; } /// Returns "best known" trip count for the specified loop \p L as defined by /// the following procedure: /// 1) Returns exact trip count if it is known. /// 2) Returns expected trip count according to profile data if any. /// 3) Returns upper bound estimate if it is known. /// 4) Returns std::nullopt if all of the above failed. static std::optional getSmallBestKnownTC(ScalarEvolution &SE, Loop *L) { // Check if exact trip count is known. if (unsigned ExpectedTC = SE.getSmallConstantTripCount(L)) return ExpectedTC; // Check if there is an expected trip count available from profile data. if (LoopVectorizeWithBlockFrequency) if (auto EstimatedTC = getLoopEstimatedTripCount(L)) return *EstimatedTC; // Check if upper bound estimate is known. if (unsigned ExpectedTC = SE.getSmallConstantMaxTripCount(L)) return ExpectedTC; return std::nullopt; } /// Return a vector containing interleaved elements from multiple /// smaller input vectors. static Value *interleaveVectors(IRBuilderBase &Builder, ArrayRef Vals, const Twine &Name) { unsigned Factor = Vals.size(); assert(Factor > 1 && "Tried to interleave invalid number of vectors"); VectorType *VecTy = cast(Vals[0]->getType()); #ifndef NDEBUG for (Value *Val : Vals) assert(Val->getType() == VecTy && "Tried to interleave mismatched types"); #endif // Scalable vectors cannot use arbitrary shufflevectors (only splats), so // must use intrinsics to interleave. if (VecTy->isScalableTy()) { VectorType *WideVecTy = VectorType::getDoubleElementsVectorType(VecTy); return Builder.CreateIntrinsic( WideVecTy, Intrinsic::experimental_vector_interleave2, Vals, /*FMFSource=*/nullptr, Name); } // Fixed length. Start by concatenating all vectors into a wide vector. Value *WideVec = concatenateVectors(Builder, Vals); // Interleave the elements into the wide vector. const unsigned NumElts = VecTy->getElementCount().getFixedValue(); return Builder.CreateShuffleVector( WideVec, createInterleaveMask(NumElts, Factor), Name); } namespace { // Forward declare GeneratedRTChecks. class GeneratedRTChecks; using SCEV2ValueTy = DenseMap; } // namespace namespace llvm { AnalysisKey ShouldRunExtraVectorPasses::Key; /// InnerLoopVectorizer vectorizes loops which contain only one basic /// block to a specified vectorization factor (VF). /// This class performs the widening of scalars into vectors, or multiple /// scalars. This class also implements the following features: /// * It inserts an epilogue loop for handling loops that don't have iteration /// counts that are known to be a multiple of the vectorization factor. /// * It handles the code generation for reduction variables. /// * Scalarization (implementation using scalars) of un-vectorizable /// instructions. /// InnerLoopVectorizer does not perform any vectorization-legality /// checks, and relies on the caller to check for the different legality /// aspects. The InnerLoopVectorizer relies on the /// LoopVectorizationLegality class to provide information about the induction /// and reduction variables that were found to a given vectorization factor. class InnerLoopVectorizer { public: InnerLoopVectorizer(Loop *OrigLoop, PredicatedScalarEvolution &PSE, LoopInfo *LI, DominatorTree *DT, const TargetLibraryInfo *TLI, const TargetTransformInfo *TTI, AssumptionCache *AC, OptimizationRemarkEmitter *ORE, ElementCount VecWidth, ElementCount MinProfitableTripCount, unsigned UnrollFactor, LoopVectorizationLegality *LVL, LoopVectorizationCostModel *CM, BlockFrequencyInfo *BFI, ProfileSummaryInfo *PSI, GeneratedRTChecks &RTChecks) : OrigLoop(OrigLoop), PSE(PSE), LI(LI), DT(DT), TLI(TLI), TTI(TTI), AC(AC), ORE(ORE), VF(VecWidth), UF(UnrollFactor), Builder(PSE.getSE()->getContext()), Legal(LVL), Cost(CM), BFI(BFI), PSI(PSI), RTChecks(RTChecks) { // Query this against the original loop and save it here because the profile // of the original loop header may change as the transformation happens. OptForSizeBasedOnProfile = llvm::shouldOptimizeForSize( OrigLoop->getHeader(), PSI, BFI, PGSOQueryType::IRPass); if (MinProfitableTripCount.isZero()) this->MinProfitableTripCount = VecWidth; else this->MinProfitableTripCount = MinProfitableTripCount; } virtual ~InnerLoopVectorizer() = default; /// Create a new empty loop that will contain vectorized instructions later /// on, while the old loop will be used as the scalar remainder. Control flow /// is generated around the vectorized (and scalar epilogue) loops consisting /// of various checks and bypasses. Return the pre-header block of the new /// loop and the start value for the canonical induction, if it is != 0. The /// latter is the case when vectorizing the epilogue loop. In the case of /// epilogue vectorization, this function is overriden to handle the more /// complex control flow around the loops. \p ExpandedSCEVs is used to /// look up SCEV expansions for expressions needed during skeleton creation. virtual std::pair createVectorizedLoopSkeleton(const SCEV2ValueTy &ExpandedSCEVs); /// Fix the vectorized code, taking care of header phi's, live-outs, and more. void fixVectorizedLoop(VPTransformState &State, VPlan &Plan); // Return true if any runtime check is added. bool areSafetyChecksAdded() { return AddedSafetyChecks; } /// A type for vectorized values in the new loop. Each value from the /// original loop, when vectorized, is represented by UF vector values in the /// new unrolled loop, where UF is the unroll factor. using VectorParts = SmallVector; /// A helper function to scalarize a single Instruction in the innermost loop. /// Generates a sequence of scalar instances for each lane between \p MinLane /// and \p MaxLane, times each part between \p MinPart and \p MaxPart, /// inclusive. Uses the VPValue operands from \p RepRecipe instead of \p /// Instr's operands. void scalarizeInstruction(const Instruction *Instr, VPReplicateRecipe *RepRecipe, const VPIteration &Instance, VPTransformState &State); /// Try to vectorize interleaved access group \p Group with the base address /// given in \p Addr, optionally masking the vector operations if \p /// BlockInMask is non-null. Use \p State to translate given VPValues to IR /// values in the vectorized loop. void vectorizeInterleaveGroup(const InterleaveGroup *Group, ArrayRef VPDefs, VPTransformState &State, VPValue *Addr, ArrayRef StoredValues, VPValue *BlockInMask, bool NeedsMaskForGaps); /// Fix the non-induction PHIs in \p Plan. void fixNonInductionPHIs(VPlan &Plan, VPTransformState &State); /// Returns true if the reordering of FP operations is not allowed, but we are /// able to vectorize with strict in-order reductions for the given RdxDesc. bool useOrderedReductions(const RecurrenceDescriptor &RdxDesc); /// Create a new phi node for the induction variable \p OrigPhi to resume /// iteration count in the scalar epilogue, from where the vectorized loop /// left off. \p Step is the SCEV-expanded induction step to use. In cases /// where the loop skeleton is more complicated (i.e., epilogue vectorization) /// and the resume values can come from an additional bypass block, the \p /// AdditionalBypass pair provides information about the bypass block and the /// end value on the edge from bypass to this loop. PHINode *createInductionResumeValue( PHINode *OrigPhi, const InductionDescriptor &ID, Value *Step, ArrayRef BypassBlocks, std::pair AdditionalBypass = {nullptr, nullptr}); /// Returns the original loop trip count. Value *getTripCount() const { return TripCount; } /// Used to set the trip count after ILV's construction and after the /// preheader block has been executed. Note that this always holds the trip /// count of the original loop for both main loop and epilogue vectorization. void setTripCount(Value *TC) { TripCount = TC; } protected: friend class LoopVectorizationPlanner; /// A small list of PHINodes. using PhiVector = SmallVector; /// A type for scalarized values in the new loop. Each value from the /// original loop, when scalarized, is represented by UF x VF scalar values /// in the new unrolled loop, where UF is the unroll factor and VF is the /// vectorization factor. using ScalarParts = SmallVector, 2>; /// Set up the values of the IVs correctly when exiting the vector loop. void fixupIVUsers(PHINode *OrigPhi, const InductionDescriptor &II, Value *VectorTripCount, Value *EndValue, BasicBlock *MiddleBlock, BasicBlock *VectorHeader, VPlan &Plan, VPTransformState &State); /// Create the exit value of first order recurrences in the middle block and /// update their users. void fixFixedOrderRecurrence(VPFirstOrderRecurrencePHIRecipe *PhiR, VPTransformState &State); /// Create code for the loop exit value of the reduction. void fixReduction(VPReductionPHIRecipe *Phi, VPTransformState &State); /// Iteratively sink the scalarized operands of a predicated instruction into /// the block that was created for it. void sinkScalarOperands(Instruction *PredInst); /// Returns (and creates if needed) the trip count of the widened loop. Value *getOrCreateVectorTripCount(BasicBlock *InsertBlock); /// Returns a bitcasted value to the requested vector type. /// Also handles bitcasts of vector <-> vector types. Value *createBitOrPointerCast(Value *V, VectorType *DstVTy, const DataLayout &DL); /// Emit a bypass check to see if the vector trip count is zero, including if /// it overflows. void emitIterationCountCheck(BasicBlock *Bypass); /// Emit a bypass check to see if all of the SCEV assumptions we've /// had to make are correct. Returns the block containing the checks or /// nullptr if no checks have been added. BasicBlock *emitSCEVChecks(BasicBlock *Bypass); /// Emit bypass checks to check any memory assumptions we may have made. /// Returns the block containing the checks or nullptr if no checks have been /// added. BasicBlock *emitMemRuntimeChecks(BasicBlock *Bypass); /// Emit basic blocks (prefixed with \p Prefix) for the iteration check, /// vector loop preheader, middle block and scalar preheader. void createVectorLoopSkeleton(StringRef Prefix); /// Create new phi nodes for the induction variables to resume iteration count /// in the scalar epilogue, from where the vectorized loop left off. /// In cases where the loop skeleton is more complicated (eg. epilogue /// vectorization) and the resume values can come from an additional bypass /// block, the \p AdditionalBypass pair provides information about the bypass /// block and the end value on the edge from bypass to this loop. void createInductionResumeValues( const SCEV2ValueTy &ExpandedSCEVs, std::pair AdditionalBypass = {nullptr, nullptr}); /// Complete the loop skeleton by adding debug MDs, creating appropriate /// conditional branches in the middle block, preparing the builder and /// running the verifier. Return the preheader of the completed vector loop. BasicBlock *completeLoopSkeleton(); /// Collect poison-generating recipes that may generate a poison value that is /// used after vectorization, even when their operands are not poison. Those /// recipes meet the following conditions: /// * Contribute to the address computation of a recipe generating a widen /// memory load/store (VPWidenMemoryInstructionRecipe or /// VPInterleaveRecipe). /// * Such a widen memory load/store has at least one underlying Instruction /// that is in a basic block that needs predication and after vectorization /// the generated instruction won't be predicated. void collectPoisonGeneratingRecipes(VPTransformState &State); /// Allow subclasses to override and print debug traces before/after vplan /// execution, when trace information is requested. virtual void printDebugTracesAtStart(){}; virtual void printDebugTracesAtEnd(){}; /// The original loop. Loop *OrigLoop; /// A wrapper around ScalarEvolution used to add runtime SCEV checks. Applies /// dynamic knowledge to simplify SCEV expressions and converts them to a /// more usable form. PredicatedScalarEvolution &PSE; /// Loop Info. LoopInfo *LI; /// Dominator Tree. DominatorTree *DT; /// Target Library Info. const TargetLibraryInfo *TLI; /// Target Transform Info. const TargetTransformInfo *TTI; /// Assumption Cache. AssumptionCache *AC; /// Interface to emit optimization remarks. OptimizationRemarkEmitter *ORE; /// The vectorization SIMD factor to use. Each vector will have this many /// vector elements. ElementCount VF; ElementCount MinProfitableTripCount; /// The vectorization unroll factor to use. Each scalar is vectorized to this /// many different vector instructions. unsigned UF; /// The builder that we use IRBuilder<> Builder; // --- Vectorization state --- /// The vector-loop preheader. BasicBlock *LoopVectorPreHeader; /// The scalar-loop preheader. BasicBlock *LoopScalarPreHeader; /// Middle Block between the vector and the scalar. BasicBlock *LoopMiddleBlock; /// The unique ExitBlock of the scalar loop if one exists. Note that /// there can be multiple exiting edges reaching this block. BasicBlock *LoopExitBlock; /// The scalar loop body. BasicBlock *LoopScalarBody; /// A list of all bypass blocks. The first block is the entry of the loop. SmallVector LoopBypassBlocks; /// Store instructions that were predicated. SmallVector PredicatedInstructions; /// Trip count of the original loop. Value *TripCount = nullptr; /// Trip count of the widened loop (TripCount - TripCount % (VF*UF)) Value *VectorTripCount = nullptr; /// The legality analysis. LoopVectorizationLegality *Legal; /// The profitablity analysis. LoopVectorizationCostModel *Cost; // Record whether runtime checks are added. bool AddedSafetyChecks = false; // Holds the end values for each induction variable. We save the end values // so we can later fix-up the external users of the induction variables. DenseMap IVEndValues; /// BFI and PSI are used to check for profile guided size optimizations. BlockFrequencyInfo *BFI; ProfileSummaryInfo *PSI; // Whether this loop should be optimized for size based on profile guided size // optimizatios. bool OptForSizeBasedOnProfile; /// Structure to hold information about generated runtime checks, responsible /// for cleaning the checks, if vectorization turns out unprofitable. GeneratedRTChecks &RTChecks; // Holds the resume values for reductions in the loops, used to set the // correct start value of reduction PHIs when vectorizing the epilogue. SmallMapVector ReductionResumeValues; }; class InnerLoopUnroller : public InnerLoopVectorizer { public: InnerLoopUnroller(Loop *OrigLoop, PredicatedScalarEvolution &PSE, LoopInfo *LI, DominatorTree *DT, const TargetLibraryInfo *TLI, const TargetTransformInfo *TTI, AssumptionCache *AC, OptimizationRemarkEmitter *ORE, unsigned UnrollFactor, LoopVectorizationLegality *LVL, LoopVectorizationCostModel *CM, BlockFrequencyInfo *BFI, ProfileSummaryInfo *PSI, GeneratedRTChecks &Check) : InnerLoopVectorizer(OrigLoop, PSE, LI, DT, TLI, TTI, AC, ORE, ElementCount::getFixed(1), ElementCount::getFixed(1), UnrollFactor, LVL, CM, BFI, PSI, Check) {} }; /// Encapsulate information regarding vectorization of a loop and its epilogue. /// This information is meant to be updated and used across two stages of /// epilogue vectorization. struct EpilogueLoopVectorizationInfo { ElementCount MainLoopVF = ElementCount::getFixed(0); unsigned MainLoopUF = 0; ElementCount EpilogueVF = ElementCount::getFixed(0); unsigned EpilogueUF = 0; BasicBlock *MainLoopIterationCountCheck = nullptr; BasicBlock *EpilogueIterationCountCheck = nullptr; BasicBlock *SCEVSafetyCheck = nullptr; BasicBlock *MemSafetyCheck = nullptr; Value *TripCount = nullptr; Value *VectorTripCount = nullptr; EpilogueLoopVectorizationInfo(ElementCount MVF, unsigned MUF, ElementCount EVF, unsigned EUF) : MainLoopVF(MVF), MainLoopUF(MUF), EpilogueVF(EVF), EpilogueUF(EUF) { assert(EUF == 1 && "A high UF for the epilogue loop is likely not beneficial."); } }; /// An extension of the inner loop vectorizer that creates a skeleton for a /// vectorized loop that has its epilogue (residual) also vectorized. /// The idea is to run the vplan on a given loop twice, firstly to setup the /// skeleton and vectorize the main loop, and secondly to complete the skeleton /// from the first step and vectorize the epilogue. This is achieved by /// deriving two concrete strategy classes from this base class and invoking /// them in succession from the loop vectorizer planner. class InnerLoopAndEpilogueVectorizer : public InnerLoopVectorizer { public: InnerLoopAndEpilogueVectorizer( Loop *OrigLoop, PredicatedScalarEvolution &PSE, LoopInfo *LI, DominatorTree *DT, const TargetLibraryInfo *TLI, const TargetTransformInfo *TTI, AssumptionCache *AC, OptimizationRemarkEmitter *ORE, EpilogueLoopVectorizationInfo &EPI, LoopVectorizationLegality *LVL, llvm::LoopVectorizationCostModel *CM, BlockFrequencyInfo *BFI, ProfileSummaryInfo *PSI, GeneratedRTChecks &Checks) : InnerLoopVectorizer(OrigLoop, PSE, LI, DT, TLI, TTI, AC, ORE, EPI.MainLoopVF, EPI.MainLoopVF, EPI.MainLoopUF, LVL, CM, BFI, PSI, Checks), EPI(EPI) {} // Override this function to handle the more complex control flow around the // three loops. std::pair createVectorizedLoopSkeleton( const SCEV2ValueTy &ExpandedSCEVs) final { return createEpilogueVectorizedLoopSkeleton(ExpandedSCEVs); } /// The interface for creating a vectorized skeleton using one of two /// different strategies, each corresponding to one execution of the vplan /// as described above. virtual std::pair createEpilogueVectorizedLoopSkeleton(const SCEV2ValueTy &ExpandedSCEVs) = 0; /// Holds and updates state information required to vectorize the main loop /// and its epilogue in two separate passes. This setup helps us avoid /// regenerating and recomputing runtime safety checks. It also helps us to /// shorten the iteration-count-check path length for the cases where the /// iteration count of the loop is so small that the main vector loop is /// completely skipped. EpilogueLoopVectorizationInfo &EPI; }; /// A specialized derived class of inner loop vectorizer that performs /// vectorization of *main* loops in the process of vectorizing loops and their /// epilogues. class EpilogueVectorizerMainLoop : public InnerLoopAndEpilogueVectorizer { public: EpilogueVectorizerMainLoop( Loop *OrigLoop, PredicatedScalarEvolution &PSE, LoopInfo *LI, DominatorTree *DT, const TargetLibraryInfo *TLI, const TargetTransformInfo *TTI, AssumptionCache *AC, OptimizationRemarkEmitter *ORE, EpilogueLoopVectorizationInfo &EPI, LoopVectorizationLegality *LVL, llvm::LoopVectorizationCostModel *CM, BlockFrequencyInfo *BFI, ProfileSummaryInfo *PSI, GeneratedRTChecks &Check) : InnerLoopAndEpilogueVectorizer(OrigLoop, PSE, LI, DT, TLI, TTI, AC, ORE, EPI, LVL, CM, BFI, PSI, Check) {} /// Implements the interface for creating a vectorized skeleton using the /// *main loop* strategy (ie the first pass of vplan execution). std::pair createEpilogueVectorizedLoopSkeleton(const SCEV2ValueTy &ExpandedSCEVs) final; protected: /// Emits an iteration count bypass check once for the main loop (when \p /// ForEpilogue is false) and once for the epilogue loop (when \p /// ForEpilogue is true). BasicBlock *emitIterationCountCheck(BasicBlock *Bypass, bool ForEpilogue); void printDebugTracesAtStart() override; void printDebugTracesAtEnd() override; }; // A specialized derived class of inner loop vectorizer that performs // vectorization of *epilogue* loops in the process of vectorizing loops and // their epilogues. class EpilogueVectorizerEpilogueLoop : public InnerLoopAndEpilogueVectorizer { public: EpilogueVectorizerEpilogueLoop( Loop *OrigLoop, PredicatedScalarEvolution &PSE, LoopInfo *LI, DominatorTree *DT, const TargetLibraryInfo *TLI, const TargetTransformInfo *TTI, AssumptionCache *AC, OptimizationRemarkEmitter *ORE, EpilogueLoopVectorizationInfo &EPI, LoopVectorizationLegality *LVL, llvm::LoopVectorizationCostModel *CM, BlockFrequencyInfo *BFI, ProfileSummaryInfo *PSI, GeneratedRTChecks &Checks) : InnerLoopAndEpilogueVectorizer(OrigLoop, PSE, LI, DT, TLI, TTI, AC, ORE, EPI, LVL, CM, BFI, PSI, Checks) { TripCount = EPI.TripCount; } /// Implements the interface for creating a vectorized skeleton using the /// *epilogue loop* strategy (ie the second pass of vplan execution). std::pair createEpilogueVectorizedLoopSkeleton(const SCEV2ValueTy &ExpandedSCEVs) final; protected: /// Emits an iteration count bypass check after the main vector loop has /// finished to see if there are any iterations left to execute by either /// the vector epilogue or the scalar epilogue. BasicBlock *emitMinimumVectorEpilogueIterCountCheck( BasicBlock *Bypass, BasicBlock *Insert); void printDebugTracesAtStart() override; void printDebugTracesAtEnd() override; }; } // end namespace llvm /// Look for a meaningful debug location on the instruction or it's /// operands. static DebugLoc getDebugLocFromInstOrOperands(Instruction *I) { if (!I) return DebugLoc(); DebugLoc Empty; if (I->getDebugLoc() != Empty) return I->getDebugLoc(); for (Use &Op : I->operands()) { if (Instruction *OpInst = dyn_cast(Op)) if (OpInst->getDebugLoc() != Empty) return OpInst->getDebugLoc(); } return I->getDebugLoc(); } /// Write a \p DebugMsg about vectorization to the debug output stream. If \p I /// is passed, the message relates to that particular instruction. #ifndef NDEBUG static void debugVectorizationMessage(const StringRef Prefix, const StringRef DebugMsg, Instruction *I) { dbgs() << "LV: " << Prefix << DebugMsg; if (I != nullptr) dbgs() << " " << *I; else dbgs() << '.'; dbgs() << '\n'; } #endif /// Create an analysis remark that explains why vectorization failed /// /// \p PassName is the name of the pass (e.g. can be AlwaysPrint). \p /// RemarkName is the identifier for the remark. If \p I is passed it is an /// instruction that prevents vectorization. Otherwise \p TheLoop is used for /// the location of the remark. \return the remark object that can be /// streamed to. static OptimizationRemarkAnalysis createLVAnalysis(const char *PassName, StringRef RemarkName, Loop *TheLoop, Instruction *I) { Value *CodeRegion = TheLoop->getHeader(); DebugLoc DL = TheLoop->getStartLoc(); if (I) { CodeRegion = I->getParent(); // If there is no debug location attached to the instruction, revert back to // using the loop's. if (I->getDebugLoc()) DL = I->getDebugLoc(); } return OptimizationRemarkAnalysis(PassName, RemarkName, DL, CodeRegion); } namespace llvm { /// Return a value for Step multiplied by VF. Value *createStepForVF(IRBuilderBase &B, Type *Ty, ElementCount VF, int64_t Step) { assert(Ty->isIntegerTy() && "Expected an integer step"); return B.CreateElementCount(Ty, VF.multiplyCoefficientBy(Step)); } /// Return the runtime value for VF. Value *getRuntimeVF(IRBuilderBase &B, Type *Ty, ElementCount VF) { return B.CreateElementCount(Ty, VF); } const SCEV *createTripCountSCEV(Type *IdxTy, PredicatedScalarEvolution &PSE, Loop *OrigLoop) { const SCEV *BackedgeTakenCount = PSE.getBackedgeTakenCount(); assert(!isa(BackedgeTakenCount) && "Invalid loop count"); ScalarEvolution &SE = *PSE.getSE(); return SE.getTripCountFromExitCount(BackedgeTakenCount, IdxTy, OrigLoop); } void reportVectorizationFailure(const StringRef DebugMsg, const StringRef OREMsg, const StringRef ORETag, OptimizationRemarkEmitter *ORE, Loop *TheLoop, Instruction *I) { LLVM_DEBUG(debugVectorizationMessage("Not vectorizing: ", DebugMsg, I)); LoopVectorizeHints Hints(TheLoop, true /* doesn't matter */, *ORE); ORE->emit( createLVAnalysis(Hints.vectorizeAnalysisPassName(), ORETag, TheLoop, I) << "loop not vectorized: " << OREMsg); } void reportVectorizationInfo(const StringRef Msg, const StringRef ORETag, OptimizationRemarkEmitter *ORE, Loop *TheLoop, Instruction *I) { LLVM_DEBUG(debugVectorizationMessage("", Msg, I)); LoopVectorizeHints Hints(TheLoop, true /* doesn't matter */, *ORE); ORE->emit( createLVAnalysis(Hints.vectorizeAnalysisPassName(), ORETag, TheLoop, I) << Msg); } /// Report successful vectorization of the loop. In case an outer loop is /// vectorized, prepend "outer" to the vectorization remark. static void reportVectorization(OptimizationRemarkEmitter *ORE, Loop *TheLoop, VectorizationFactor VF, unsigned IC) { LLVM_DEBUG(debugVectorizationMessage( "Vectorizing: ", TheLoop->isInnermost() ? "innermost loop" : "outer loop", nullptr)); StringRef LoopType = TheLoop->isInnermost() ? "" : "outer "; ORE->emit([&]() { return OptimizationRemark(LV_NAME, "Vectorized", TheLoop->getStartLoc(), TheLoop->getHeader()) << "vectorized " << LoopType << "loop (vectorization width: " << ore::NV("VectorizationFactor", VF.Width) << ", interleaved count: " << ore::NV("InterleaveCount", IC) << ")"; }); } } // end namespace llvm #ifndef NDEBUG /// \return string containing a file name and a line # for the given loop. static std::string getDebugLocString(const Loop *L) { std::string Result; if (L) { raw_string_ostream OS(Result); if (const DebugLoc LoopDbgLoc = L->getStartLoc()) LoopDbgLoc.print(OS); else // Just print the module name. OS << L->getHeader()->getParent()->getParent()->getModuleIdentifier(); OS.flush(); } return Result; } #endif void InnerLoopVectorizer::collectPoisonGeneratingRecipes( VPTransformState &State) { // Collect recipes in the backward slice of `Root` that may generate a poison // value that is used after vectorization. SmallPtrSet Visited; auto collectPoisonGeneratingInstrsInBackwardSlice([&](VPRecipeBase *Root) { SmallVector Worklist; Worklist.push_back(Root); // Traverse the backward slice of Root through its use-def chain. while (!Worklist.empty()) { VPRecipeBase *CurRec = Worklist.back(); Worklist.pop_back(); if (!Visited.insert(CurRec).second) continue; // Prune search if we find another recipe generating a widen memory // instruction. Widen memory instructions involved in address computation // will lead to gather/scatter instructions, which don't need to be // handled. if (isa(CurRec) || isa(CurRec) || isa(CurRec) || isa(CurRec) || isa(CurRec)) continue; // This recipe contributes to the address computation of a widen // load/store. If the underlying instruction has poison-generating flags, // drop them directly. if (auto *RecWithFlags = dyn_cast(CurRec)) { RecWithFlags->dropPoisonGeneratingFlags(); } else { Instruction *Instr = dyn_cast_or_null( CurRec->getVPSingleValue()->getUnderlyingValue()); (void)Instr; assert((!Instr || !Instr->hasPoisonGeneratingFlags()) && "found instruction with poison generating flags not covered by " "VPRecipeWithIRFlags"); } // Add new definitions to the worklist. for (VPValue *operand : CurRec->operands()) if (VPRecipeBase *OpDef = operand->getDefiningRecipe()) Worklist.push_back(OpDef); } }); // Traverse all the recipes in the VPlan and collect the poison-generating // recipes in the backward slice starting at the address of a VPWidenRecipe or // VPInterleaveRecipe. auto Iter = vp_depth_first_deep(State.Plan->getEntry()); for (VPBasicBlock *VPBB : VPBlockUtils::blocksOnly(Iter)) { for (VPRecipeBase &Recipe : *VPBB) { if (auto *WidenRec = dyn_cast(&Recipe)) { Instruction &UnderlyingInstr = WidenRec->getIngredient(); VPRecipeBase *AddrDef = WidenRec->getAddr()->getDefiningRecipe(); if (AddrDef && WidenRec->isConsecutive() && Legal->blockNeedsPredication(UnderlyingInstr.getParent())) collectPoisonGeneratingInstrsInBackwardSlice(AddrDef); } else if (auto *InterleaveRec = dyn_cast(&Recipe)) { VPRecipeBase *AddrDef = InterleaveRec->getAddr()->getDefiningRecipe(); if (AddrDef) { // Check if any member of the interleave group needs predication. const InterleaveGroup *InterGroup = InterleaveRec->getInterleaveGroup(); bool NeedPredication = false; for (int I = 0, NumMembers = InterGroup->getNumMembers(); I < NumMembers; ++I) { Instruction *Member = InterGroup->getMember(I); if (Member) NeedPredication |= Legal->blockNeedsPredication(Member->getParent()); } if (NeedPredication) collectPoisonGeneratingInstrsInBackwardSlice(AddrDef); } } } } } namespace llvm { // Loop vectorization cost-model hints how the scalar epilogue loop should be // lowered. enum ScalarEpilogueLowering { // The default: allowing scalar epilogues. CM_ScalarEpilogueAllowed, // Vectorization with OptForSize: don't allow epilogues. CM_ScalarEpilogueNotAllowedOptSize, // A special case of vectorisation with OptForSize: loops with a very small // trip count are considered for vectorization under OptForSize, thereby // making sure the cost of their loop body is dominant, free of runtime // guards and scalar iteration overheads. CM_ScalarEpilogueNotAllowedLowTripLoop, // Loop hint predicate indicating an epilogue is undesired. CM_ScalarEpilogueNotNeededUsePredicate, // Directive indicating we must either tail fold or not vectorize CM_ScalarEpilogueNotAllowedUsePredicate }; using InstructionVFPair = std::pair; /// LoopVectorizationCostModel - estimates the expected speedups due to /// vectorization. /// In many cases vectorization is not profitable. This can happen because of /// a number of reasons. In this class we mainly attempt to predict the /// expected speedup/slowdowns due to the supported instruction set. We use the /// TargetTransformInfo to query the different backends for the cost of /// different operations. class LoopVectorizationCostModel { public: LoopVectorizationCostModel(ScalarEpilogueLowering SEL, Loop *L, PredicatedScalarEvolution &PSE, LoopInfo *LI, LoopVectorizationLegality *Legal, const TargetTransformInfo &TTI, const TargetLibraryInfo *TLI, DemandedBits *DB, AssumptionCache *AC, OptimizationRemarkEmitter *ORE, const Function *F, const LoopVectorizeHints *Hints, InterleavedAccessInfo &IAI) : ScalarEpilogueStatus(SEL), TheLoop(L), PSE(PSE), LI(LI), Legal(Legal), TTI(TTI), TLI(TLI), DB(DB), AC(AC), ORE(ORE), TheFunction(F), Hints(Hints), InterleaveInfo(IAI) {} /// \return An upper bound for the vectorization factors (both fixed and /// scalable). If the factors are 0, vectorization and interleaving should be /// avoided up front. FixedScalableVFPair computeMaxVF(ElementCount UserVF, unsigned UserIC); /// \return True if runtime checks are required for vectorization, and false /// otherwise. bool runtimeChecksRequired(); /// Setup cost-based decisions for user vectorization factor. /// \return true if the UserVF is a feasible VF to be chosen. bool selectUserVectorizationFactor(ElementCount UserVF) { collectUniformsAndScalars(UserVF); collectInstsToScalarize(UserVF); return expectedCost(UserVF).first.isValid(); } /// \return The size (in bits) of the smallest and widest types in the code /// that needs to be vectorized. We ignore values that remain scalar such as /// 64 bit loop indices. std::pair getSmallestAndWidestTypes(); /// \return The desired interleave count. /// If interleave count has been specified by metadata it will be returned. /// Otherwise, the interleave count is computed and returned. VF and LoopCost /// are the selected vectorization factor and the cost of the selected VF. unsigned selectInterleaveCount(ElementCount VF, InstructionCost LoopCost); /// Memory access instruction may be vectorized in more than one way. /// Form of instruction after vectorization depends on cost. /// This function takes cost-based decisions for Load/Store instructions /// and collects them in a map. This decisions map is used for building /// the lists of loop-uniform and loop-scalar instructions. /// The calculated cost is saved with widening decision in order to /// avoid redundant calculations. void setCostBasedWideningDecision(ElementCount VF); /// A call may be vectorized in different ways depending on whether we have /// vectorized variants available and whether the target supports masking. /// This function analyzes all calls in the function at the supplied VF, /// makes a decision based on the costs of available options, and stores that /// decision in a map for use in planning and plan execution. void setVectorizedCallDecision(ElementCount VF); /// A struct that represents some properties of the register usage /// of a loop. struct RegisterUsage { /// Holds the number of loop invariant values that are used in the loop. /// The key is ClassID of target-provided register class. SmallMapVector LoopInvariantRegs; /// Holds the maximum number of concurrent live intervals in the loop. /// The key is ClassID of target-provided register class. SmallMapVector MaxLocalUsers; }; /// \return Returns information about the register usages of the loop for the /// given vectorization factors. SmallVector calculateRegisterUsage(ArrayRef VFs); /// Collect values we want to ignore in the cost model. void collectValuesToIgnore(); /// Collect all element types in the loop for which widening is needed. void collectElementTypesForWidening(); /// Split reductions into those that happen in the loop, and those that happen /// outside. In loop reductions are collected into InLoopReductions. void collectInLoopReductions(); /// Returns true if we should use strict in-order reductions for the given /// RdxDesc. This is true if the -enable-strict-reductions flag is passed, /// the IsOrdered flag of RdxDesc is set and we do not allow reordering /// of FP operations. bool useOrderedReductions(const RecurrenceDescriptor &RdxDesc) const { return !Hints->allowReordering() && RdxDesc.isOrdered(); } /// \returns The smallest bitwidth each instruction can be represented with. /// The vector equivalents of these instructions should be truncated to this /// type. const MapVector &getMinimalBitwidths() const { return MinBWs; } /// \returns True if it is more profitable to scalarize instruction \p I for /// vectorization factor \p VF. bool isProfitableToScalarize(Instruction *I, ElementCount VF) const { assert(VF.isVector() && "Profitable to scalarize relevant only for VF > 1."); // Cost model is not run in the VPlan-native path - return conservative // result until this changes. if (EnableVPlanNativePath) return false; auto Scalars = InstsToScalarize.find(VF); assert(Scalars != InstsToScalarize.end() && "VF not yet analyzed for scalarization profitability"); return Scalars->second.contains(I); } /// Returns true if \p I is known to be uniform after vectorization. bool isUniformAfterVectorization(Instruction *I, ElementCount VF) const { // Pseudo probe needs to be duplicated for each unrolled iteration and // vector lane so that profiled loop trip count can be accurately // accumulated instead of being under counted. if (isa(I)) return false; if (VF.isScalar()) return true; // Cost model is not run in the VPlan-native path - return conservative // result until this changes. if (EnableVPlanNativePath) return false; auto UniformsPerVF = Uniforms.find(VF); assert(UniformsPerVF != Uniforms.end() && "VF not yet analyzed for uniformity"); return UniformsPerVF->second.count(I); } /// Returns true if \p I is known to be scalar after vectorization. bool isScalarAfterVectorization(Instruction *I, ElementCount VF) const { if (VF.isScalar()) return true; // Cost model is not run in the VPlan-native path - return conservative // result until this changes. if (EnableVPlanNativePath) return false; auto ScalarsPerVF = Scalars.find(VF); assert(ScalarsPerVF != Scalars.end() && "Scalar values are not calculated for VF"); return ScalarsPerVF->second.count(I); } /// \returns True if instruction \p I can be truncated to a smaller bitwidth /// for vectorization factor \p VF. bool canTruncateToMinimalBitwidth(Instruction *I, ElementCount VF) const { return VF.isVector() && MinBWs.contains(I) && !isProfitableToScalarize(I, VF) && !isScalarAfterVectorization(I, VF); } /// Decision that was taken during cost calculation for memory instruction. enum InstWidening { CM_Unknown, CM_Widen, // For consecutive accesses with stride +1. CM_Widen_Reverse, // For consecutive accesses with stride -1. CM_Interleave, CM_GatherScatter, CM_Scalarize, CM_VectorCall, CM_IntrinsicCall }; /// Save vectorization decision \p W and \p Cost taken by the cost model for /// instruction \p I and vector width \p VF. void setWideningDecision(Instruction *I, ElementCount VF, InstWidening W, InstructionCost Cost) { assert(VF.isVector() && "Expected VF >=2"); WideningDecisions[std::make_pair(I, VF)] = std::make_pair(W, Cost); } /// Save vectorization decision \p W and \p Cost taken by the cost model for /// interleaving group \p Grp and vector width \p VF. void setWideningDecision(const InterleaveGroup *Grp, ElementCount VF, InstWidening W, InstructionCost Cost) { assert(VF.isVector() && "Expected VF >=2"); /// Broadcast this decicion to all instructions inside the group. /// But the cost will be assigned to one instruction only. for (unsigned i = 0; i < Grp->getFactor(); ++i) { if (auto *I = Grp->getMember(i)) { if (Grp->getInsertPos() == I) WideningDecisions[std::make_pair(I, VF)] = std::make_pair(W, Cost); else WideningDecisions[std::make_pair(I, VF)] = std::make_pair(W, 0); } } } /// Return the cost model decision for the given instruction \p I and vector /// width \p VF. Return CM_Unknown if this instruction did not pass /// through the cost modeling. InstWidening getWideningDecision(Instruction *I, ElementCount VF) const { assert(VF.isVector() && "Expected VF to be a vector VF"); // Cost model is not run in the VPlan-native path - return conservative // result until this changes. if (EnableVPlanNativePath) return CM_GatherScatter; std::pair InstOnVF = std::make_pair(I, VF); auto Itr = WideningDecisions.find(InstOnVF); if (Itr == WideningDecisions.end()) return CM_Unknown; return Itr->second.first; } /// Return the vectorization cost for the given instruction \p I and vector /// width \p VF. InstructionCost getWideningCost(Instruction *I, ElementCount VF) { assert(VF.isVector() && "Expected VF >=2"); std::pair InstOnVF = std::make_pair(I, VF); assert(WideningDecisions.contains(InstOnVF) && "The cost is not calculated"); return WideningDecisions[InstOnVF].second; } struct CallWideningDecision { InstWidening Kind; Function *Variant; Intrinsic::ID IID; std::optional MaskPos; InstructionCost Cost; }; void setCallWideningDecision(CallInst *CI, ElementCount VF, InstWidening Kind, Function *Variant, Intrinsic::ID IID, std::optional MaskPos, InstructionCost Cost) { assert(!VF.isScalar() && "Expected vector VF"); CallWideningDecisions[std::make_pair(CI, VF)] = {Kind, Variant, IID, MaskPos, Cost}; } CallWideningDecision getCallWideningDecision(CallInst *CI, ElementCount VF) const { assert(!VF.isScalar() && "Expected vector VF"); return CallWideningDecisions.at(std::make_pair(CI, VF)); } /// Return True if instruction \p I is an optimizable truncate whose operand /// is an induction variable. Such a truncate will be removed by adding a new /// induction variable with the destination type. bool isOptimizableIVTruncate(Instruction *I, ElementCount VF) { // If the instruction is not a truncate, return false. auto *Trunc = dyn_cast(I); if (!Trunc) return false; // Get the source and destination types of the truncate. Type *SrcTy = ToVectorTy(cast(I)->getSrcTy(), VF); Type *DestTy = ToVectorTy(cast(I)->getDestTy(), VF); // If the truncate is free for the given types, return false. Replacing a // free truncate with an induction variable would add an induction variable // update instruction to each iteration of the loop. We exclude from this // check the primary induction variable since it will need an update // instruction regardless. Value *Op = Trunc->getOperand(0); if (Op != Legal->getPrimaryInduction() && TTI.isTruncateFree(SrcTy, DestTy)) return false; // If the truncated value is not an induction variable, return false. return Legal->isInductionPhi(Op); } /// Collects the instructions to scalarize for each predicated instruction in /// the loop. void collectInstsToScalarize(ElementCount VF); /// Collect Uniform and Scalar values for the given \p VF. /// The sets depend on CM decision for Load/Store instructions /// that may be vectorized as interleave, gather-scatter or scalarized. /// Also make a decision on what to do about call instructions in the loop /// at that VF -- scalarize, call a known vector routine, or call a /// vector intrinsic. void collectUniformsAndScalars(ElementCount VF) { // Do the analysis once. if (VF.isScalar() || Uniforms.contains(VF)) return; setCostBasedWideningDecision(VF); setVectorizedCallDecision(VF); collectLoopUniforms(VF); collectLoopScalars(VF); } /// Returns true if the target machine supports masked store operation /// for the given \p DataType and kind of access to \p Ptr. bool isLegalMaskedStore(Type *DataType, Value *Ptr, Align Alignment) const { return Legal->isConsecutivePtr(DataType, Ptr) && TTI.isLegalMaskedStore(DataType, Alignment); } /// Returns true if the target machine supports masked load operation /// for the given \p DataType and kind of access to \p Ptr. bool isLegalMaskedLoad(Type *DataType, Value *Ptr, Align Alignment) const { return Legal->isConsecutivePtr(DataType, Ptr) && TTI.isLegalMaskedLoad(DataType, Alignment); } /// Returns true if the target machine can represent \p V as a masked gather /// or scatter operation. bool isLegalGatherOrScatter(Value *V, ElementCount VF) { bool LI = isa(V); bool SI = isa(V); if (!LI && !SI) return false; auto *Ty = getLoadStoreType(V); Align Align = getLoadStoreAlignment(V); if (VF.isVector()) Ty = VectorType::get(Ty, VF); return (LI && TTI.isLegalMaskedGather(Ty, Align)) || (SI && TTI.isLegalMaskedScatter(Ty, Align)); } /// Returns true if the target machine supports all of the reduction /// variables found for the given VF. bool canVectorizeReductions(ElementCount VF) const { return (all_of(Legal->getReductionVars(), [&](auto &Reduction) -> bool { const RecurrenceDescriptor &RdxDesc = Reduction.second; return TTI.isLegalToVectorizeReduction(RdxDesc, VF); })); } /// Given costs for both strategies, return true if the scalar predication /// lowering should be used for div/rem. This incorporates an override /// option so it is not simply a cost comparison. bool isDivRemScalarWithPredication(InstructionCost ScalarCost, InstructionCost SafeDivisorCost) const { switch (ForceSafeDivisor) { case cl::BOU_UNSET: return ScalarCost < SafeDivisorCost; case cl::BOU_TRUE: return false; case cl::BOU_FALSE: return true; }; llvm_unreachable("impossible case value"); } /// Returns true if \p I is an instruction which requires predication and /// for which our chosen predication strategy is scalarization (i.e. we /// don't have an alternate strategy such as masking available). /// \p VF is the vectorization factor that will be used to vectorize \p I. bool isScalarWithPredication(Instruction *I, ElementCount VF) const; /// Returns true if \p I is an instruction that needs to be predicated /// at runtime. The result is independent of the predication mechanism. /// Superset of instructions that return true for isScalarWithPredication. bool isPredicatedInst(Instruction *I) const; /// Return the costs for our two available strategies for lowering a /// div/rem operation which requires speculating at least one lane. /// First result is for scalarization (will be invalid for scalable /// vectors); second is for the safe-divisor strategy. std::pair getDivRemSpeculationCost(Instruction *I, ElementCount VF) const; /// Returns true if \p I is a memory instruction with consecutive memory /// access that can be widened. bool memoryInstructionCanBeWidened(Instruction *I, ElementCount VF); /// Returns true if \p I is a memory instruction in an interleaved-group /// of memory accesses that can be vectorized with wide vector loads/stores /// and shuffles. bool interleavedAccessCanBeWidened(Instruction *I, ElementCount VF); /// Check if \p Instr belongs to any interleaved access group. bool isAccessInterleaved(Instruction *Instr) { return InterleaveInfo.isInterleaved(Instr); } /// Get the interleaved access group that \p Instr belongs to. const InterleaveGroup * getInterleavedAccessGroup(Instruction *Instr) { return InterleaveInfo.getInterleaveGroup(Instr); } /// Returns true if we're required to use a scalar epilogue for at least /// the final iteration of the original loop. bool requiresScalarEpilogue(bool IsVectorizing) const { if (!isScalarEpilogueAllowed()) return false; // If we might exit from anywhere but the latch, must run the exiting // iteration in scalar form. if (TheLoop->getExitingBlock() != TheLoop->getLoopLatch()) return true; return IsVectorizing && InterleaveInfo.requiresScalarEpilogue(); } /// Returns true if we're required to use a scalar epilogue for at least /// the final iteration of the original loop for all VFs in \p Range. /// A scalar epilogue must either be required for all VFs in \p Range or for /// none. bool requiresScalarEpilogue(VFRange Range) const { auto RequiresScalarEpilogue = [this](ElementCount VF) { return requiresScalarEpilogue(VF.isVector()); }; bool IsRequired = all_of(Range, RequiresScalarEpilogue); assert( (IsRequired || none_of(Range, RequiresScalarEpilogue)) && "all VFs in range must agree on whether a scalar epilogue is required"); return IsRequired; } /// Returns true if a scalar epilogue is not allowed due to optsize or a /// loop hint annotation. bool isScalarEpilogueAllowed() const { return ScalarEpilogueStatus == CM_ScalarEpilogueAllowed; } /// Returns the TailFoldingStyle that is best for the current loop. TailFoldingStyle getTailFoldingStyle(bool IVUpdateMayOverflow = true) const { if (!CanFoldTailByMasking) return TailFoldingStyle::None; if (ForceTailFoldingStyle.getNumOccurrences()) return ForceTailFoldingStyle; return TTI.getPreferredTailFoldingStyle(IVUpdateMayOverflow); } /// Returns true if all loop blocks should be masked to fold tail loop. bool foldTailByMasking() const { return getTailFoldingStyle() != TailFoldingStyle::None; } /// Returns true if the instructions in this block requires predication /// for any reason, e.g. because tail folding now requires a predicate /// or because the block in the original loop was predicated. bool blockNeedsPredicationForAnyReason(BasicBlock *BB) const { return foldTailByMasking() || Legal->blockNeedsPredication(BB); } /// Returns true if the Phi is part of an inloop reduction. bool isInLoopReduction(PHINode *Phi) const { return InLoopReductions.contains(Phi); } /// Estimate cost of an intrinsic call instruction CI if it were vectorized /// with factor VF. Return the cost of the instruction, including /// scalarization overhead if it's needed. InstructionCost getVectorIntrinsicCost(CallInst *CI, ElementCount VF) const; /// Estimate cost of a call instruction CI if it were vectorized with factor /// VF. Return the cost of the instruction, including scalarization overhead /// if it's needed. InstructionCost getVectorCallCost(CallInst *CI, ElementCount VF) const; /// Invalidates decisions already taken by the cost model. void invalidateCostModelingDecisions() { WideningDecisions.clear(); CallWideningDecisions.clear(); Uniforms.clear(); Scalars.clear(); } /// The vectorization cost is a combination of the cost itself and a boolean /// indicating whether any of the contributing operations will actually /// operate on vector values after type legalization in the backend. If this /// latter value is false, then all operations will be scalarized (i.e. no /// vectorization has actually taken place). using VectorizationCostTy = std::pair; /// Returns the expected execution cost. The unit of the cost does /// not matter because we use the 'cost' units to compare different /// vector widths. The cost that is returned is *not* normalized by /// the factor width. If \p Invalid is not nullptr, this function /// will add a pair(Instruction*, ElementCount) to \p Invalid for /// each instruction that has an Invalid cost for the given VF. VectorizationCostTy expectedCost(ElementCount VF, SmallVectorImpl *Invalid = nullptr); bool hasPredStores() const { return NumPredStores > 0; } /// Returns true if epilogue vectorization is considered profitable, and /// false otherwise. /// \p VF is the vectorization factor chosen for the original loop. bool isEpilogueVectorizationProfitable(const ElementCount VF) const; private: unsigned NumPredStores = 0; /// \return An upper bound for the vectorization factors for both /// fixed and scalable vectorization, where the minimum-known number of /// elements is a power-of-2 larger than zero. If scalable vectorization is /// disabled or unsupported, then the scalable part will be equal to /// ElementCount::getScalable(0). FixedScalableVFPair computeFeasibleMaxVF(unsigned MaxTripCount, ElementCount UserVF, bool FoldTailByMasking); /// \return the maximized element count based on the targets vector /// registers and the loop trip-count, but limited to a maximum safe VF. /// This is a helper function of computeFeasibleMaxVF. ElementCount getMaximizedVFForTarget(unsigned MaxTripCount, unsigned SmallestType, unsigned WidestType, ElementCount MaxSafeVF, bool FoldTailByMasking); /// \return the maximum legal scalable VF, based on the safe max number /// of elements. ElementCount getMaxLegalScalableVF(unsigned MaxSafeElements); /// Returns the execution time cost of an instruction for a given vector /// width. Vector width of one means scalar. VectorizationCostTy getInstructionCost(Instruction *I, ElementCount VF); /// The cost-computation logic from getInstructionCost which provides /// the vector type as an output parameter. InstructionCost getInstructionCost(Instruction *I, ElementCount VF, Type *&VectorTy); /// Return the cost of instructions in an inloop reduction pattern, if I is /// part of that pattern. std::optional getReductionPatternCost(Instruction *I, ElementCount VF, Type *VectorTy, TTI::TargetCostKind CostKind) const; /// Calculate vectorization cost of memory instruction \p I. InstructionCost getMemoryInstructionCost(Instruction *I, ElementCount VF); /// The cost computation for scalarized memory instruction. InstructionCost getMemInstScalarizationCost(Instruction *I, ElementCount VF); /// The cost computation for interleaving group of memory instructions. InstructionCost getInterleaveGroupCost(Instruction *I, ElementCount VF); /// The cost computation for Gather/Scatter instruction. InstructionCost getGatherScatterCost(Instruction *I, ElementCount VF); /// The cost computation for widening instruction \p I with consecutive /// memory access. InstructionCost getConsecutiveMemOpCost(Instruction *I, ElementCount VF); /// The cost calculation for Load/Store instruction \p I with uniform pointer - /// Load: scalar load + broadcast. /// Store: scalar store + (loop invariant value stored? 0 : extract of last /// element) InstructionCost getUniformMemOpCost(Instruction *I, ElementCount VF); /// Estimate the overhead of scalarizing an instruction. This is a /// convenience wrapper for the type-based getScalarizationOverhead API. InstructionCost getScalarizationOverhead(Instruction *I, ElementCount VF, TTI::TargetCostKind CostKind) const; /// Returns true if an artificially high cost for emulated masked memrefs /// should be used. bool useEmulatedMaskMemRefHack(Instruction *I, ElementCount VF); /// Map of scalar integer values to the smallest bitwidth they can be legally /// represented as. The vector equivalents of these values should be truncated /// to this type. MapVector MinBWs; /// A type representing the costs for instructions if they were to be /// scalarized rather than vectorized. The entries are Instruction-Cost /// pairs. using ScalarCostsTy = DenseMap; /// A set containing all BasicBlocks that are known to present after /// vectorization as a predicated block. DenseMap> PredicatedBBsAfterVectorization; /// Records whether it is allowed to have the original scalar loop execute at /// least once. This may be needed as a fallback loop in case runtime /// aliasing/dependence checks fail, or to handle the tail/remainder /// iterations when the trip count is unknown or doesn't divide by the VF, /// or as a peel-loop to handle gaps in interleave-groups. /// Under optsize and when the trip count is very small we don't allow any /// iterations to execute in the scalar loop. ScalarEpilogueLowering ScalarEpilogueStatus = CM_ScalarEpilogueAllowed; /// All blocks of loop are to be masked to fold tail of scalar iterations. bool CanFoldTailByMasking = false; /// A map holding scalar costs for different vectorization factors. The /// presence of a cost for an instruction in the mapping indicates that the /// instruction will be scalarized when vectorizing with the associated /// vectorization factor. The entries are VF-ScalarCostTy pairs. DenseMap InstsToScalarize; /// Holds the instructions known to be uniform after vectorization. /// The data is collected per VF. DenseMap> Uniforms; /// Holds the instructions known to be scalar after vectorization. /// The data is collected per VF. DenseMap> Scalars; /// Holds the instructions (address computations) that are forced to be /// scalarized. DenseMap> ForcedScalars; /// PHINodes of the reductions that should be expanded in-loop. SmallPtrSet InLoopReductions; /// A Map of inloop reduction operations and their immediate chain operand. /// FIXME: This can be removed once reductions can be costed correctly in /// VPlan. This was added to allow quick lookup of the inloop operations. DenseMap InLoopReductionImmediateChains; /// Returns the expected difference in cost from scalarizing the expression /// feeding a predicated instruction \p PredInst. The instructions to /// scalarize and their scalar costs are collected in \p ScalarCosts. A /// non-negative return value implies the expression will be scalarized. /// Currently, only single-use chains are considered for scalarization. InstructionCost computePredInstDiscount(Instruction *PredInst, ScalarCostsTy &ScalarCosts, ElementCount VF); /// Collect the instructions that are uniform after vectorization. An /// instruction is uniform if we represent it with a single scalar value in /// the vectorized loop corresponding to each vector iteration. Examples of /// uniform instructions include pointer operands of consecutive or /// interleaved memory accesses. Note that although uniformity implies an /// instruction will be scalar, the reverse is not true. In general, a /// scalarized instruction will be represented by VF scalar values in the /// vectorized loop, each corresponding to an iteration of the original /// scalar loop. void collectLoopUniforms(ElementCount VF); /// Collect the instructions that are scalar after vectorization. An /// instruction is scalar if it is known to be uniform or will be scalarized /// during vectorization. collectLoopScalars should only add non-uniform nodes /// to the list if they are used by a load/store instruction that is marked as /// CM_Scalarize. Non-uniform scalarized instructions will be represented by /// VF values in the vectorized loop, each corresponding to an iteration of /// the original scalar loop. void collectLoopScalars(ElementCount VF); /// Keeps cost model vectorization decision and cost for instructions. /// Right now it is used for memory instructions only. using DecisionList = DenseMap, std::pair>; DecisionList WideningDecisions; using CallDecisionList = DenseMap, CallWideningDecision>; CallDecisionList CallWideningDecisions; /// Returns true if \p V is expected to be vectorized and it needs to be /// extracted. bool needsExtract(Value *V, ElementCount VF) const { Instruction *I = dyn_cast(V); if (VF.isScalar() || !I || !TheLoop->contains(I) || TheLoop->isLoopInvariant(I)) return false; // Assume we can vectorize V (and hence we need extraction) if the // scalars are not computed yet. This can happen, because it is called // via getScalarizationOverhead from setCostBasedWideningDecision, before // the scalars are collected. That should be a safe assumption in most // cases, because we check if the operands have vectorizable types // beforehand in LoopVectorizationLegality. return !Scalars.contains(VF) || !isScalarAfterVectorization(I, VF); }; /// Returns a range containing only operands needing to be extracted. SmallVector filterExtractingOperands(Instruction::op_range Ops, ElementCount VF) const { return SmallVector(make_filter_range( Ops, [this, VF](Value *V) { return this->needsExtract(V, VF); })); } public: /// The loop that we evaluate. Loop *TheLoop; /// Predicated scalar evolution analysis. PredicatedScalarEvolution &PSE; /// Loop Info analysis. LoopInfo *LI; /// Vectorization legality. LoopVectorizationLegality *Legal; /// Vector target information. const TargetTransformInfo &TTI; /// Target Library Info. const TargetLibraryInfo *TLI; /// Demanded bits analysis. DemandedBits *DB; /// Assumption cache. AssumptionCache *AC; /// Interface to emit optimization remarks. OptimizationRemarkEmitter *ORE; const Function *TheFunction; /// Loop Vectorize Hint. const LoopVectorizeHints *Hints; /// The interleave access information contains groups of interleaved accesses /// with the same stride and close to each other. InterleavedAccessInfo &InterleaveInfo; /// Values to ignore in the cost model. SmallPtrSet ValuesToIgnore; /// Values to ignore in the cost model when VF > 1. SmallPtrSet VecValuesToIgnore; /// All element types found in the loop. SmallPtrSet ElementTypesInLoop; }; } // end namespace llvm namespace { /// Helper struct to manage generating runtime checks for vectorization. /// /// The runtime checks are created up-front in temporary blocks to allow better /// estimating the cost and un-linked from the existing IR. After deciding to /// vectorize, the checks are moved back. If deciding not to vectorize, the /// temporary blocks are completely removed. class GeneratedRTChecks { /// Basic block which contains the generated SCEV checks, if any. BasicBlock *SCEVCheckBlock = nullptr; /// The value representing the result of the generated SCEV checks. If it is /// nullptr, either no SCEV checks have been generated or they have been used. Value *SCEVCheckCond = nullptr; /// Basic block which contains the generated memory runtime checks, if any. BasicBlock *MemCheckBlock = nullptr; /// The value representing the result of the generated memory runtime checks. /// If it is nullptr, either no memory runtime checks have been generated or /// they have been used. Value *MemRuntimeCheckCond = nullptr; DominatorTree *DT; LoopInfo *LI; TargetTransformInfo *TTI; SCEVExpander SCEVExp; SCEVExpander MemCheckExp; bool CostTooHigh = false; const bool AddBranchWeights; Loop *OuterLoop = nullptr; public: GeneratedRTChecks(ScalarEvolution &SE, DominatorTree *DT, LoopInfo *LI, TargetTransformInfo *TTI, const DataLayout &DL, bool AddBranchWeights) : DT(DT), LI(LI), TTI(TTI), SCEVExp(SE, DL, "scev.check"), MemCheckExp(SE, DL, "scev.check"), AddBranchWeights(AddBranchWeights) {} /// Generate runtime checks in SCEVCheckBlock and MemCheckBlock, so we can /// accurately estimate the cost of the runtime checks. The blocks are /// un-linked from the IR and is added back during vector code generation. If /// there is no vector code generation, the check blocks are removed /// completely. void Create(Loop *L, const LoopAccessInfo &LAI, const SCEVPredicate &UnionPred, ElementCount VF, unsigned IC) { // Hard cutoff to limit compile-time increase in case a very large number of // runtime checks needs to be generated. // TODO: Skip cutoff if the loop is guaranteed to execute, e.g. due to // profile info. CostTooHigh = LAI.getNumRuntimePointerChecks() > VectorizeMemoryCheckThreshold; if (CostTooHigh) return; BasicBlock *LoopHeader = L->getHeader(); BasicBlock *Preheader = L->getLoopPreheader(); // Use SplitBlock to create blocks for SCEV & memory runtime checks to // ensure the blocks are properly added to LoopInfo & DominatorTree. Those // may be used by SCEVExpander. The blocks will be un-linked from their // predecessors and removed from LI & DT at the end of the function. if (!UnionPred.isAlwaysTrue()) { SCEVCheckBlock = SplitBlock(Preheader, Preheader->getTerminator(), DT, LI, nullptr, "vector.scevcheck"); SCEVCheckCond = SCEVExp.expandCodeForPredicate( &UnionPred, SCEVCheckBlock->getTerminator()); } const auto &RtPtrChecking = *LAI.getRuntimePointerChecking(); if (RtPtrChecking.Need) { auto *Pred = SCEVCheckBlock ? SCEVCheckBlock : Preheader; MemCheckBlock = SplitBlock(Pred, Pred->getTerminator(), DT, LI, nullptr, "vector.memcheck"); auto DiffChecks = RtPtrChecking.getDiffChecks(); if (DiffChecks) { Value *RuntimeVF = nullptr; MemRuntimeCheckCond = addDiffRuntimeChecks( MemCheckBlock->getTerminator(), *DiffChecks, MemCheckExp, [VF, &RuntimeVF](IRBuilderBase &B, unsigned Bits) { if (!RuntimeVF) RuntimeVF = getRuntimeVF(B, B.getIntNTy(Bits), VF); return RuntimeVF; }, IC); } else { MemRuntimeCheckCond = addRuntimeChecks( MemCheckBlock->getTerminator(), L, RtPtrChecking.getChecks(), MemCheckExp, VectorizerParams::HoistRuntimeChecks); } assert(MemRuntimeCheckCond && "no RT checks generated although RtPtrChecking " "claimed checks are required"); } if (!MemCheckBlock && !SCEVCheckBlock) return; // Unhook the temporary block with the checks, update various places // accordingly. if (SCEVCheckBlock) SCEVCheckBlock->replaceAllUsesWith(Preheader); if (MemCheckBlock) MemCheckBlock->replaceAllUsesWith(Preheader); if (SCEVCheckBlock) { SCEVCheckBlock->getTerminator()->moveBefore(Preheader->getTerminator()); new UnreachableInst(Preheader->getContext(), SCEVCheckBlock); Preheader->getTerminator()->eraseFromParent(); } if (MemCheckBlock) { MemCheckBlock->getTerminator()->moveBefore(Preheader->getTerminator()); new UnreachableInst(Preheader->getContext(), MemCheckBlock); Preheader->getTerminator()->eraseFromParent(); } DT->changeImmediateDominator(LoopHeader, Preheader); if (MemCheckBlock) { DT->eraseNode(MemCheckBlock); LI->removeBlock(MemCheckBlock); } if (SCEVCheckBlock) { DT->eraseNode(SCEVCheckBlock); LI->removeBlock(SCEVCheckBlock); } // Outer loop is used as part of the later cost calculations. OuterLoop = L->getParentLoop(); } InstructionCost getCost() { if (SCEVCheckBlock || MemCheckBlock) LLVM_DEBUG(dbgs() << "Calculating cost of runtime checks:\n"); if (CostTooHigh) { InstructionCost Cost; Cost.setInvalid(); LLVM_DEBUG(dbgs() << " number of checks exceeded threshold\n"); return Cost; } InstructionCost RTCheckCost = 0; if (SCEVCheckBlock) for (Instruction &I : *SCEVCheckBlock) { if (SCEVCheckBlock->getTerminator() == &I) continue; InstructionCost C = TTI->getInstructionCost(&I, TTI::TCK_RecipThroughput); LLVM_DEBUG(dbgs() << " " << C << " for " << I << "\n"); RTCheckCost += C; } if (MemCheckBlock) { InstructionCost MemCheckCost = 0; for (Instruction &I : *MemCheckBlock) { if (MemCheckBlock->getTerminator() == &I) continue; InstructionCost C = TTI->getInstructionCost(&I, TTI::TCK_RecipThroughput); LLVM_DEBUG(dbgs() << " " << C << " for " << I << "\n"); MemCheckCost += C; } // If the runtime memory checks are being created inside an outer loop // we should find out if these checks are outer loop invariant. If so, // the checks will likely be hoisted out and so the effective cost will // reduce according to the outer loop trip count. if (OuterLoop) { ScalarEvolution *SE = MemCheckExp.getSE(); // TODO: If profitable, we could refine this further by analysing every // individual memory check, since there could be a mixture of loop // variant and invariant checks that mean the final condition is // variant. const SCEV *Cond = SE->getSCEV(MemRuntimeCheckCond); if (SE->isLoopInvariant(Cond, OuterLoop)) { // It seems reasonable to assume that we can reduce the effective // cost of the checks even when we know nothing about the trip // count. Assume that the outer loop executes at least twice. unsigned BestTripCount = 2; // If exact trip count is known use that. if (unsigned SmallTC = SE->getSmallConstantTripCount(OuterLoop)) BestTripCount = SmallTC; else if (LoopVectorizeWithBlockFrequency) { // Else use profile data if available. if (auto EstimatedTC = getLoopEstimatedTripCount(OuterLoop)) BestTripCount = *EstimatedTC; } InstructionCost NewMemCheckCost = MemCheckCost / BestTripCount; // Let's ensure the cost is always at least 1. NewMemCheckCost = std::max(*NewMemCheckCost.getValue(), (InstructionCost::CostType)1); LLVM_DEBUG(dbgs() << "We expect runtime memory checks to be hoisted " << "out of the outer loop. Cost reduced from " << MemCheckCost << " to " << NewMemCheckCost << '\n'); MemCheckCost = NewMemCheckCost; } } RTCheckCost += MemCheckCost; } if (SCEVCheckBlock || MemCheckBlock) LLVM_DEBUG(dbgs() << "Total cost of runtime checks: " << RTCheckCost << "\n"); return RTCheckCost; } /// Remove the created SCEV & memory runtime check blocks & instructions, if /// unused. ~GeneratedRTChecks() { SCEVExpanderCleaner SCEVCleaner(SCEVExp); SCEVExpanderCleaner MemCheckCleaner(MemCheckExp); if (!SCEVCheckCond) SCEVCleaner.markResultUsed(); if (!MemRuntimeCheckCond) MemCheckCleaner.markResultUsed(); if (MemRuntimeCheckCond) { auto &SE = *MemCheckExp.getSE(); // Memory runtime check generation creates compares that use expanded // values. Remove them before running the SCEVExpanderCleaners. for (auto &I : make_early_inc_range(reverse(*MemCheckBlock))) { if (MemCheckExp.isInsertedInstruction(&I)) continue; SE.forgetValue(&I); I.eraseFromParent(); } } MemCheckCleaner.cleanup(); SCEVCleaner.cleanup(); if (SCEVCheckCond) SCEVCheckBlock->eraseFromParent(); if (MemRuntimeCheckCond) MemCheckBlock->eraseFromParent(); } /// Adds the generated SCEVCheckBlock before \p LoopVectorPreHeader and /// adjusts the branches to branch to the vector preheader or \p Bypass, /// depending on the generated condition. BasicBlock *emitSCEVChecks(BasicBlock *Bypass, BasicBlock *LoopVectorPreHeader, BasicBlock *LoopExitBlock) { if (!SCEVCheckCond) return nullptr; Value *Cond = SCEVCheckCond; // Mark the check as used, to prevent it from being removed during cleanup. SCEVCheckCond = nullptr; if (auto *C = dyn_cast(Cond)) if (C->isZero()) return nullptr; auto *Pred = LoopVectorPreHeader->getSinglePredecessor(); BranchInst::Create(LoopVectorPreHeader, SCEVCheckBlock); // Create new preheader for vector loop. if (OuterLoop) OuterLoop->addBasicBlockToLoop(SCEVCheckBlock, *LI); SCEVCheckBlock->getTerminator()->eraseFromParent(); SCEVCheckBlock->moveBefore(LoopVectorPreHeader); Pred->getTerminator()->replaceSuccessorWith(LoopVectorPreHeader, SCEVCheckBlock); DT->addNewBlock(SCEVCheckBlock, Pred); DT->changeImmediateDominator(LoopVectorPreHeader, SCEVCheckBlock); BranchInst &BI = *BranchInst::Create(Bypass, LoopVectorPreHeader, Cond); if (AddBranchWeights) setBranchWeights(BI, SCEVCheckBypassWeights); ReplaceInstWithInst(SCEVCheckBlock->getTerminator(), &BI); return SCEVCheckBlock; } /// Adds the generated MemCheckBlock before \p LoopVectorPreHeader and adjusts /// the branches to branch to the vector preheader or \p Bypass, depending on /// the generated condition. BasicBlock *emitMemRuntimeChecks(BasicBlock *Bypass, BasicBlock *LoopVectorPreHeader) { // Check if we generated code that checks in runtime if arrays overlap. if (!MemRuntimeCheckCond) return nullptr; auto *Pred = LoopVectorPreHeader->getSinglePredecessor(); Pred->getTerminator()->replaceSuccessorWith(LoopVectorPreHeader, MemCheckBlock); DT->addNewBlock(MemCheckBlock, Pred); DT->changeImmediateDominator(LoopVectorPreHeader, MemCheckBlock); MemCheckBlock->moveBefore(LoopVectorPreHeader); if (OuterLoop) OuterLoop->addBasicBlockToLoop(MemCheckBlock, *LI); BranchInst &BI = *BranchInst::Create(Bypass, LoopVectorPreHeader, MemRuntimeCheckCond); if (AddBranchWeights) { setBranchWeights(BI, MemCheckBypassWeights); } ReplaceInstWithInst(MemCheckBlock->getTerminator(), &BI); MemCheckBlock->getTerminator()->setDebugLoc( Pred->getTerminator()->getDebugLoc()); // Mark the check as used, to prevent it from being removed during cleanup. MemRuntimeCheckCond = nullptr; return MemCheckBlock; } }; } // namespace static bool useActiveLaneMask(TailFoldingStyle Style) { return Style == TailFoldingStyle::Data || Style == TailFoldingStyle::DataAndControlFlow || Style == TailFoldingStyle::DataAndControlFlowWithoutRuntimeCheck; } static bool useActiveLaneMaskForControlFlow(TailFoldingStyle Style) { return Style == TailFoldingStyle::DataAndControlFlow || Style == TailFoldingStyle::DataAndControlFlowWithoutRuntimeCheck; } // Return true if \p OuterLp is an outer loop annotated with hints for explicit // vectorization. The loop needs to be annotated with #pragma omp simd // simdlen(#) or #pragma clang vectorize(enable) vectorize_width(#). If the // vector length information is not provided, vectorization is not considered // explicit. Interleave hints are not allowed either. These limitations will be // relaxed in the future. // Please, note that we are currently forced to abuse the pragma 'clang // vectorize' semantics. This pragma provides *auto-vectorization hints* // (i.e., LV must check that vectorization is legal) whereas pragma 'omp simd' // provides *explicit vectorization hints* (LV can bypass legal checks and // assume that vectorization is legal). However, both hints are implemented // using the same metadata (llvm.loop.vectorize, processed by // LoopVectorizeHints). This will be fixed in the future when the native IR // representation for pragma 'omp simd' is introduced. static bool isExplicitVecOuterLoop(Loop *OuterLp, OptimizationRemarkEmitter *ORE) { assert(!OuterLp->isInnermost() && "This is not an outer loop"); LoopVectorizeHints Hints(OuterLp, true /*DisableInterleaving*/, *ORE); // Only outer loops with an explicit vectorization hint are supported. // Unannotated outer loops are ignored. if (Hints.getForce() == LoopVectorizeHints::FK_Undefined) return false; Function *Fn = OuterLp->getHeader()->getParent(); if (!Hints.allowVectorization(Fn, OuterLp, true /*VectorizeOnlyWhenForced*/)) { LLVM_DEBUG(dbgs() << "LV: Loop hints prevent outer loop vectorization.\n"); return false; } if (Hints.getInterleave() > 1) { // TODO: Interleave support is future work. LLVM_DEBUG(dbgs() << "LV: Not vectorizing: Interleave is not supported for " "outer loops.\n"); Hints.emitRemarkWithHints(); return false; } return true; } static void collectSupportedLoops(Loop &L, LoopInfo *LI, OptimizationRemarkEmitter *ORE, SmallVectorImpl &V) { // Collect inner loops and outer loops without irreducible control flow. For // now, only collect outer loops that have explicit vectorization hints. If we // are stress testing the VPlan H-CFG construction, we collect the outermost // loop of every loop nest. if (L.isInnermost() || VPlanBuildStressTest || (EnableVPlanNativePath && isExplicitVecOuterLoop(&L, ORE))) { LoopBlocksRPO RPOT(&L); RPOT.perform(LI); if (!containsIrreducibleCFG(RPOT, *LI)) { V.push_back(&L); // TODO: Collect inner loops inside marked outer loops in case // vectorization fails for the outer loop. Do not invoke // 'containsIrreducibleCFG' again for inner loops when the outer loop is // already known to be reducible. We can use an inherited attribute for // that. return; } } for (Loop *InnerL : L) collectSupportedLoops(*InnerL, LI, ORE, V); } //===----------------------------------------------------------------------===// // Implementation of LoopVectorizationLegality, InnerLoopVectorizer and // LoopVectorizationCostModel and LoopVectorizationPlanner. //===----------------------------------------------------------------------===// /// Compute the transformed value of Index at offset StartValue using step /// StepValue. /// For integer induction, returns StartValue + Index * StepValue. /// For pointer induction, returns StartValue[Index * StepValue]. /// FIXME: The newly created binary instructions should contain nsw/nuw /// flags, which can be found from the original scalar operations. static Value * emitTransformedIndex(IRBuilderBase &B, Value *Index, Value *StartValue, Value *Step, InductionDescriptor::InductionKind InductionKind, const BinaryOperator *InductionBinOp) { Type *StepTy = Step->getType(); Value *CastedIndex = StepTy->isIntegerTy() ? B.CreateSExtOrTrunc(Index, StepTy) : B.CreateCast(Instruction::SIToFP, Index, StepTy); if (CastedIndex != Index) { CastedIndex->setName(CastedIndex->getName() + ".cast"); Index = CastedIndex; } // Note: the IR at this point is broken. We cannot use SE to create any new // SCEV and then expand it, hoping that SCEV's simplification will give us // a more optimal code. Unfortunately, attempt of doing so on invalid IR may // lead to various SCEV crashes. So all we can do is to use builder and rely // on InstCombine for future simplifications. Here we handle some trivial // cases only. auto CreateAdd = [&B](Value *X, Value *Y) { assert(X->getType() == Y->getType() && "Types don't match!"); if (auto *CX = dyn_cast(X)) if (CX->isZero()) return Y; if (auto *CY = dyn_cast(Y)) if (CY->isZero()) return X; return B.CreateAdd(X, Y); }; // We allow X to be a vector type, in which case Y will potentially be // splatted into a vector with the same element count. auto CreateMul = [&B](Value *X, Value *Y) { assert(X->getType()->getScalarType() == Y->getType() && "Types don't match!"); if (auto *CX = dyn_cast(X)) if (CX->isOne()) return Y; if (auto *CY = dyn_cast(Y)) if (CY->isOne()) return X; VectorType *XVTy = dyn_cast(X->getType()); if (XVTy && !isa(Y->getType())) Y = B.CreateVectorSplat(XVTy->getElementCount(), Y); return B.CreateMul(X, Y); }; switch (InductionKind) { case InductionDescriptor::IK_IntInduction: { assert(!isa(Index->getType()) && "Vector indices not supported for integer inductions yet"); assert(Index->getType() == StartValue->getType() && "Index type does not match StartValue type"); if (isa(Step) && cast(Step)->isMinusOne()) return B.CreateSub(StartValue, Index); auto *Offset = CreateMul(Index, Step); return CreateAdd(StartValue, Offset); } case InductionDescriptor::IK_PtrInduction: return B.CreatePtrAdd(StartValue, CreateMul(Index, Step)); case InductionDescriptor::IK_FpInduction: { assert(!isa(Index->getType()) && "Vector indices not supported for FP inductions yet"); assert(Step->getType()->isFloatingPointTy() && "Expected FP Step value"); assert(InductionBinOp && (InductionBinOp->getOpcode() == Instruction::FAdd || InductionBinOp->getOpcode() == Instruction::FSub) && "Original bin op should be defined for FP induction"); Value *MulExp = B.CreateFMul(Step, Index); return B.CreateBinOp(InductionBinOp->getOpcode(), StartValue, MulExp, "induction"); } case InductionDescriptor::IK_NoInduction: return nullptr; } llvm_unreachable("invalid enum"); } std::optional getMaxVScale(const Function &F, const TargetTransformInfo &TTI) { if (std::optional MaxVScale = TTI.getMaxVScale()) return MaxVScale; if (F.hasFnAttribute(Attribute::VScaleRange)) return F.getFnAttribute(Attribute::VScaleRange).getVScaleRangeMax(); return std::nullopt; } /// For the given VF and UF and maximum trip count computed for the loop, return /// whether the induction variable might overflow in the vectorized loop. If not, /// then we know a runtime overflow check always evaluates to false and can be /// removed. static bool isIndvarOverflowCheckKnownFalse( const LoopVectorizationCostModel *Cost, ElementCount VF, std::optional UF = std::nullopt) { // Always be conservative if we don't know the exact unroll factor. unsigned MaxUF = UF ? *UF : Cost->TTI.getMaxInterleaveFactor(VF); Type *IdxTy = Cost->Legal->getWidestInductionType(); APInt MaxUIntTripCount = cast(IdxTy)->getMask(); // We know the runtime overflow check is known false iff the (max) trip-count // is known and (max) trip-count + (VF * UF) does not overflow in the type of // the vector loop induction variable. if (unsigned TC = Cost->PSE.getSE()->getSmallConstantMaxTripCount(Cost->TheLoop)) { uint64_t MaxVF = VF.getKnownMinValue(); if (VF.isScalable()) { std::optional MaxVScale = getMaxVScale(*Cost->TheFunction, Cost->TTI); if (!MaxVScale) return false; MaxVF *= *MaxVScale; } return (MaxUIntTripCount - TC).ugt(MaxVF * MaxUF); } return false; } // Return whether we allow using masked interleave-groups (for dealing with // strided loads/stores that reside in predicated blocks, or for dealing // with gaps). static bool useMaskedInterleavedAccesses(const TargetTransformInfo &TTI) { // If an override option has been passed in for interleaved accesses, use it. if (EnableMaskedInterleavedMemAccesses.getNumOccurrences() > 0) return EnableMaskedInterleavedMemAccesses; return TTI.enableMaskedInterleavedAccessVectorization(); } // Try to vectorize the interleave group that \p Instr belongs to. // // E.g. Translate following interleaved load group (factor = 3): // for (i = 0; i < N; i+=3) { // R = Pic[i]; // Member of index 0 // G = Pic[i+1]; // Member of index 1 // B = Pic[i+2]; // Member of index 2 // ... // do something to R, G, B // } // To: // %wide.vec = load <12 x i32> ; Read 4 tuples of R,G,B // %R.vec = shuffle %wide.vec, poison, <0, 3, 6, 9> ; R elements // %G.vec = shuffle %wide.vec, poison, <1, 4, 7, 10> ; G elements // %B.vec = shuffle %wide.vec, poison, <2, 5, 8, 11> ; B elements // // Or translate following interleaved store group (factor = 3): // for (i = 0; i < N; i+=3) { // ... do something to R, G, B // Pic[i] = R; // Member of index 0 // Pic[i+1] = G; // Member of index 1 // Pic[i+2] = B; // Member of index 2 // } // To: // %R_G.vec = shuffle %R.vec, %G.vec, <0, 1, 2, ..., 7> // %B_U.vec = shuffle %B.vec, poison, <0, 1, 2, 3, u, u, u, u> // %interleaved.vec = shuffle %R_G.vec, %B_U.vec, // <0, 4, 8, 1, 5, 9, 2, 6, 10, 3, 7, 11> ; Interleave R,G,B elements // store <12 x i32> %interleaved.vec ; Write 4 tuples of R,G,B void InnerLoopVectorizer::vectorizeInterleaveGroup( const InterleaveGroup *Group, ArrayRef VPDefs, VPTransformState &State, VPValue *Addr, ArrayRef StoredValues, VPValue *BlockInMask, bool NeedsMaskForGaps) { Instruction *Instr = Group->getInsertPos(); const DataLayout &DL = Instr->getModule()->getDataLayout(); // Prepare for the vector type of the interleaved load/store. Type *ScalarTy = getLoadStoreType(Instr); unsigned InterleaveFactor = Group->getFactor(); auto *VecTy = VectorType::get(ScalarTy, VF * InterleaveFactor); // Prepare for the new pointers. SmallVector AddrParts; unsigned Index = Group->getIndex(Instr); // TODO: extend the masked interleaved-group support to reversed access. assert((!BlockInMask || !Group->isReverse()) && "Reversed masked interleave-group not supported."); Value *Idx; // If the group is reverse, adjust the index to refer to the last vector lane // instead of the first. We adjust the index from the first vector lane, // rather than directly getting the pointer for lane VF - 1, because the // pointer operand of the interleaved access is supposed to be uniform. For // uniform instructions, we're only required to generate a value for the // first vector lane in each unroll iteration. if (Group->isReverse()) { Value *RuntimeVF = getRuntimeVF(Builder, Builder.getInt32Ty(), VF); Idx = Builder.CreateSub(RuntimeVF, Builder.getInt32(1)); Idx = Builder.CreateMul(Idx, Builder.getInt32(Group->getFactor())); Idx = Builder.CreateAdd(Idx, Builder.getInt32(Index)); Idx = Builder.CreateNeg(Idx); } else Idx = Builder.getInt32(-Index); for (unsigned Part = 0; Part < UF; Part++) { Value *AddrPart = State.get(Addr, VPIteration(Part, 0)); if (auto *I = dyn_cast(AddrPart)) State.setDebugLocFrom(I->getDebugLoc()); // Notice current instruction could be any index. Need to adjust the address // to the member of index 0. // // E.g. a = A[i+1]; // Member of index 1 (Current instruction) // b = A[i]; // Member of index 0 // Current pointer is pointed to A[i+1], adjust it to A[i]. // // E.g. A[i+1] = a; // Member of index 1 // A[i] = b; // Member of index 0 // A[i+2] = c; // Member of index 2 (Current instruction) // Current pointer is pointed to A[i+2], adjust it to A[i]. bool InBounds = false; if (auto *gep = dyn_cast(AddrPart->stripPointerCasts())) InBounds = gep->isInBounds(); AddrPart = Builder.CreateGEP(ScalarTy, AddrPart, Idx, "", InBounds); AddrParts.push_back(AddrPart); } State.setDebugLocFrom(Instr->getDebugLoc()); Value *PoisonVec = PoisonValue::get(VecTy); auto CreateGroupMask = [this, &BlockInMask, &State, &InterleaveFactor]( unsigned Part, Value *MaskForGaps) -> Value * { if (VF.isScalable()) { assert(!MaskForGaps && "Interleaved groups with gaps are not supported."); assert(InterleaveFactor == 2 && "Unsupported deinterleave factor for scalable vectors"); auto *BlockInMaskPart = State.get(BlockInMask, Part); SmallVector Ops = {BlockInMaskPart, BlockInMaskPart}; auto *MaskTy = VectorType::get(Builder.getInt1Ty(), VF.getKnownMinValue() * 2, true); return Builder.CreateIntrinsic( MaskTy, Intrinsic::experimental_vector_interleave2, Ops, /*FMFSource=*/nullptr, "interleaved.mask"); } if (!BlockInMask) return MaskForGaps; Value *BlockInMaskPart = State.get(BlockInMask, Part); Value *ShuffledMask = Builder.CreateShuffleVector( BlockInMaskPart, createReplicatedMask(InterleaveFactor, VF.getKnownMinValue()), "interleaved.mask"); return MaskForGaps ? Builder.CreateBinOp(Instruction::And, ShuffledMask, MaskForGaps) : ShuffledMask; }; // Vectorize the interleaved load group. if (isa(Instr)) { Value *MaskForGaps = nullptr; if (NeedsMaskForGaps) { MaskForGaps = createBitMaskForGaps(Builder, VF.getKnownMinValue(), *Group); assert(MaskForGaps && "Mask for Gaps is required but it is null"); } // For each unroll part, create a wide load for the group. SmallVector NewLoads; for (unsigned Part = 0; Part < UF; Part++) { Instruction *NewLoad; if (BlockInMask || MaskForGaps) { assert(useMaskedInterleavedAccesses(*TTI) && "masked interleaved groups are not allowed."); Value *GroupMask = CreateGroupMask(Part, MaskForGaps); NewLoad = Builder.CreateMaskedLoad(VecTy, AddrParts[Part], Group->getAlign(), GroupMask, PoisonVec, "wide.masked.vec"); } else NewLoad = Builder.CreateAlignedLoad(VecTy, AddrParts[Part], Group->getAlign(), "wide.vec"); Group->addMetadata(NewLoad); NewLoads.push_back(NewLoad); } if (VecTy->isScalableTy()) { assert(InterleaveFactor == 2 && "Unsupported deinterleave factor for scalable vectors"); for (unsigned Part = 0; Part < UF; ++Part) { // Scalable vectors cannot use arbitrary shufflevectors (only splats), // so must use intrinsics to deinterleave. Value *DI = Builder.CreateIntrinsic( Intrinsic::experimental_vector_deinterleave2, VecTy, NewLoads[Part], /*FMFSource=*/nullptr, "strided.vec"); unsigned J = 0; for (unsigned I = 0; I < InterleaveFactor; ++I) { Instruction *Member = Group->getMember(I); if (!Member) continue; Value *StridedVec = Builder.CreateExtractValue(DI, I); // If this member has different type, cast the result type. if (Member->getType() != ScalarTy) { VectorType *OtherVTy = VectorType::get(Member->getType(), VF); StridedVec = createBitOrPointerCast(StridedVec, OtherVTy, DL); } if (Group->isReverse()) StridedVec = Builder.CreateVectorReverse(StridedVec, "reverse"); State.set(VPDefs[J], StridedVec, Part); ++J; } } return; } // For each member in the group, shuffle out the appropriate data from the // wide loads. unsigned J = 0; for (unsigned I = 0; I < InterleaveFactor; ++I) { Instruction *Member = Group->getMember(I); // Skip the gaps in the group. if (!Member) continue; auto StrideMask = createStrideMask(I, InterleaveFactor, VF.getKnownMinValue()); for (unsigned Part = 0; Part < UF; Part++) { Value *StridedVec = Builder.CreateShuffleVector( NewLoads[Part], StrideMask, "strided.vec"); // If this member has different type, cast the result type. if (Member->getType() != ScalarTy) { assert(!VF.isScalable() && "VF is assumed to be non scalable."); VectorType *OtherVTy = VectorType::get(Member->getType(), VF); StridedVec = createBitOrPointerCast(StridedVec, OtherVTy, DL); } if (Group->isReverse()) StridedVec = Builder.CreateVectorReverse(StridedVec, "reverse"); State.set(VPDefs[J], StridedVec, Part); } ++J; } return; } // The sub vector type for current instruction. auto *SubVT = VectorType::get(ScalarTy, VF); // Vectorize the interleaved store group. Value *MaskForGaps = createBitMaskForGaps(Builder, VF.getKnownMinValue(), *Group); assert((!MaskForGaps || useMaskedInterleavedAccesses(*TTI)) && "masked interleaved groups are not allowed."); assert((!MaskForGaps || !VF.isScalable()) && "masking gaps for scalable vectors is not yet supported."); for (unsigned Part = 0; Part < UF; Part++) { // Collect the stored vector from each member. SmallVector StoredVecs; unsigned StoredIdx = 0; for (unsigned i = 0; i < InterleaveFactor; i++) { assert((Group->getMember(i) || MaskForGaps) && "Fail to get a member from an interleaved store group"); Instruction *Member = Group->getMember(i); // Skip the gaps in the group. if (!Member) { Value *Undef = PoisonValue::get(SubVT); StoredVecs.push_back(Undef); continue; } Value *StoredVec = State.get(StoredValues[StoredIdx], Part); ++StoredIdx; if (Group->isReverse()) StoredVec = Builder.CreateVectorReverse(StoredVec, "reverse"); // If this member has different type, cast it to a unified type. if (StoredVec->getType() != SubVT) StoredVec = createBitOrPointerCast(StoredVec, SubVT, DL); StoredVecs.push_back(StoredVec); } // Interleave all the smaller vectors into one wider vector. Value *IVec = interleaveVectors(Builder, StoredVecs, "interleaved.vec"); Instruction *NewStoreInstr; if (BlockInMask || MaskForGaps) { Value *GroupMask = CreateGroupMask(Part, MaskForGaps); NewStoreInstr = Builder.CreateMaskedStore(IVec, AddrParts[Part], Group->getAlign(), GroupMask); } else NewStoreInstr = Builder.CreateAlignedStore(IVec, AddrParts[Part], Group->getAlign()); Group->addMetadata(NewStoreInstr); } } void InnerLoopVectorizer::scalarizeInstruction(const Instruction *Instr, VPReplicateRecipe *RepRecipe, const VPIteration &Instance, VPTransformState &State) { assert(!Instr->getType()->isAggregateType() && "Can't handle vectors"); // llvm.experimental.noalias.scope.decl intrinsics must only be duplicated for // the first lane and part. if (isa(Instr)) if (!Instance.isFirstIteration()) return; // Does this instruction return a value ? bool IsVoidRetTy = Instr->getType()->isVoidTy(); Instruction *Cloned = Instr->clone(); if (!IsVoidRetTy) { Cloned->setName(Instr->getName() + ".cloned"); #if !defined(NDEBUG) // Verify that VPlan type inference results agree with the type of the // generated values. assert(State.TypeAnalysis.inferScalarType(RepRecipe) == Cloned->getType() && "inferred type and type from generated instructions do not match"); #endif } RepRecipe->setFlags(Cloned); if (auto DL = Instr->getDebugLoc()) State.setDebugLocFrom(DL); // Replace the operands of the cloned instructions with their scalar // equivalents in the new loop. for (const auto &I : enumerate(RepRecipe->operands())) { auto InputInstance = Instance; VPValue *Operand = I.value(); if (vputils::isUniformAfterVectorization(Operand)) InputInstance.Lane = VPLane::getFirstLane(); Cloned->setOperand(I.index(), State.get(Operand, InputInstance)); } State.addNewMetadata(Cloned, Instr); // Place the cloned scalar in the new loop. State.Builder.Insert(Cloned); State.set(RepRecipe, Cloned, Instance); // If we just cloned a new assumption, add it the assumption cache. if (auto *II = dyn_cast(Cloned)) AC->registerAssumption(II); // End if-block. bool IfPredicateInstr = RepRecipe->getParent()->getParent()->isReplicator(); if (IfPredicateInstr) PredicatedInstructions.push_back(Cloned); } Value * InnerLoopVectorizer::getOrCreateVectorTripCount(BasicBlock *InsertBlock) { if (VectorTripCount) return VectorTripCount; Value *TC = getTripCount(); IRBuilder<> Builder(InsertBlock->getTerminator()); Type *Ty = TC->getType(); // This is where we can make the step a runtime constant. Value *Step = createStepForVF(Builder, Ty, VF, UF); // If the tail is to be folded by masking, round the number of iterations N // up to a multiple of Step instead of rounding down. This is done by first // adding Step-1 and then rounding down. Note that it's ok if this addition // overflows: the vector induction variable will eventually wrap to zero given // that it starts at zero and its Step is a power of two; the loop will then // exit, with the last early-exit vector comparison also producing all-true. // For scalable vectors the VF is not guaranteed to be a power of 2, but this // is accounted for in emitIterationCountCheck that adds an overflow check. if (Cost->foldTailByMasking()) { assert(isPowerOf2_32(VF.getKnownMinValue() * UF) && "VF*UF must be a power of 2 when folding tail by masking"); Value *NumLanes = getRuntimeVF(Builder, Ty, VF * UF); TC = Builder.CreateAdd( TC, Builder.CreateSub(NumLanes, ConstantInt::get(Ty, 1)), "n.rnd.up"); } // Now we need to generate the expression for the part of the loop that the // vectorized body will execute. This is equal to N - (N % Step) if scalar // iterations are not required for correctness, or N - Step, otherwise. Step // is equal to the vectorization factor (number of SIMD elements) times the // unroll factor (number of SIMD instructions). Value *R = Builder.CreateURem(TC, Step, "n.mod.vf"); // There are cases where we *must* run at least one iteration in the remainder // loop. See the cost model for when this can happen. If the step evenly // divides the trip count, we set the remainder to be equal to the step. If // the step does not evenly divide the trip count, no adjustment is necessary // since there will already be scalar iterations. Note that the minimum // iterations check ensures that N >= Step. if (Cost->requiresScalarEpilogue(VF.isVector())) { auto *IsZero = Builder.CreateICmpEQ(R, ConstantInt::get(R->getType(), 0)); R = Builder.CreateSelect(IsZero, Step, R); } VectorTripCount = Builder.CreateSub(TC, R, "n.vec"); return VectorTripCount; } Value *InnerLoopVectorizer::createBitOrPointerCast(Value *V, VectorType *DstVTy, const DataLayout &DL) { // Verify that V is a vector type with same number of elements as DstVTy. auto *DstFVTy = cast(DstVTy); auto VF = DstFVTy->getElementCount(); auto *SrcVecTy = cast(V->getType()); assert(VF == SrcVecTy->getElementCount() && "Vector dimensions do not match"); Type *SrcElemTy = SrcVecTy->getElementType(); Type *DstElemTy = DstFVTy->getElementType(); assert((DL.getTypeSizeInBits(SrcElemTy) == DL.getTypeSizeInBits(DstElemTy)) && "Vector elements must have same size"); // Do a direct cast if element types are castable. if (CastInst::isBitOrNoopPointerCastable(SrcElemTy, DstElemTy, DL)) { return Builder.CreateBitOrPointerCast(V, DstFVTy); } // V cannot be directly casted to desired vector type. // May happen when V is a floating point vector but DstVTy is a vector of // pointers or vice-versa. Handle this using a two-step bitcast using an // intermediate Integer type for the bitcast i.e. Ptr <-> Int <-> Float. assert((DstElemTy->isPointerTy() != SrcElemTy->isPointerTy()) && "Only one type should be a pointer type"); assert((DstElemTy->isFloatingPointTy() != SrcElemTy->isFloatingPointTy()) && "Only one type should be a floating point type"); Type *IntTy = IntegerType::getIntNTy(V->getContext(), DL.getTypeSizeInBits(SrcElemTy)); auto *VecIntTy = VectorType::get(IntTy, VF); Value *CastVal = Builder.CreateBitOrPointerCast(V, VecIntTy); return Builder.CreateBitOrPointerCast(CastVal, DstFVTy); } void InnerLoopVectorizer::emitIterationCountCheck(BasicBlock *Bypass) { Value *Count = getTripCount(); // Reuse existing vector loop preheader for TC checks. // Note that new preheader block is generated for vector loop. BasicBlock *const TCCheckBlock = LoopVectorPreHeader; IRBuilder<> Builder(TCCheckBlock->getTerminator()); // Generate code to check if the loop's trip count is less than VF * UF, or // equal to it in case a scalar epilogue is required; this implies that the // vector trip count is zero. This check also covers the case where adding one // to the backedge-taken count overflowed leading to an incorrect trip count // of zero. In this case we will also jump to the scalar loop. auto P = Cost->requiresScalarEpilogue(VF.isVector()) ? ICmpInst::ICMP_ULE : ICmpInst::ICMP_ULT; // If tail is to be folded, vector loop takes care of all iterations. Type *CountTy = Count->getType(); Value *CheckMinIters = Builder.getFalse(); auto CreateStep = [&]() -> Value * { // Create step with max(MinProTripCount, UF * VF). if (UF * VF.getKnownMinValue() >= MinProfitableTripCount.getKnownMinValue()) return createStepForVF(Builder, CountTy, VF, UF); Value *MinProfTC = createStepForVF(Builder, CountTy, MinProfitableTripCount, 1); if (!VF.isScalable()) return MinProfTC; return Builder.CreateBinaryIntrinsic( Intrinsic::umax, MinProfTC, createStepForVF(Builder, CountTy, VF, UF)); }; TailFoldingStyle Style = Cost->getTailFoldingStyle(); if (Style == TailFoldingStyle::None) CheckMinIters = Builder.CreateICmp(P, Count, CreateStep(), "min.iters.check"); else if (VF.isScalable() && !isIndvarOverflowCheckKnownFalse(Cost, VF, UF) && Style != TailFoldingStyle::DataAndControlFlowWithoutRuntimeCheck) { // vscale is not necessarily a power-of-2, which means we cannot guarantee // an overflow to zero when updating induction variables and so an // additional overflow check is required before entering the vector loop. // Get the maximum unsigned value for the type. Value *MaxUIntTripCount = ConstantInt::get(CountTy, cast(CountTy)->getMask()); Value *LHS = Builder.CreateSub(MaxUIntTripCount, Count); // Don't execute the vector loop if (UMax - n) < (VF * UF). CheckMinIters = Builder.CreateICmp(ICmpInst::ICMP_ULT, LHS, CreateStep()); } // Create new preheader for vector loop. LoopVectorPreHeader = SplitBlock(TCCheckBlock, TCCheckBlock->getTerminator(), DT, LI, nullptr, "vector.ph"); assert(DT->properlyDominates(DT->getNode(TCCheckBlock), DT->getNode(Bypass)->getIDom()) && "TC check is expected to dominate Bypass"); // Update dominator for Bypass & LoopExit (if needed). DT->changeImmediateDominator(Bypass, TCCheckBlock); if (!Cost->requiresScalarEpilogue(VF.isVector())) // If there is an epilogue which must run, there's no edge from the // middle block to exit blocks and thus no need to update the immediate // dominator of the exit blocks. DT->changeImmediateDominator(LoopExitBlock, TCCheckBlock); BranchInst &BI = *BranchInst::Create(Bypass, LoopVectorPreHeader, CheckMinIters); if (hasBranchWeightMD(*OrigLoop->getLoopLatch()->getTerminator())) setBranchWeights(BI, MinItersBypassWeights); ReplaceInstWithInst(TCCheckBlock->getTerminator(), &BI); LoopBypassBlocks.push_back(TCCheckBlock); } BasicBlock *InnerLoopVectorizer::emitSCEVChecks(BasicBlock *Bypass) { BasicBlock *const SCEVCheckBlock = RTChecks.emitSCEVChecks(Bypass, LoopVectorPreHeader, LoopExitBlock); if (!SCEVCheckBlock) return nullptr; assert(!(SCEVCheckBlock->getParent()->hasOptSize() || (OptForSizeBasedOnProfile && Cost->Hints->getForce() != LoopVectorizeHints::FK_Enabled)) && "Cannot SCEV check stride or overflow when optimizing for size"); // Update dominator only if this is first RT check. if (LoopBypassBlocks.empty()) { DT->changeImmediateDominator(Bypass, SCEVCheckBlock); if (!Cost->requiresScalarEpilogue(VF.isVector())) // If there is an epilogue which must run, there's no edge from the // middle block to exit blocks and thus no need to update the immediate // dominator of the exit blocks. DT->changeImmediateDominator(LoopExitBlock, SCEVCheckBlock); } LoopBypassBlocks.push_back(SCEVCheckBlock); AddedSafetyChecks = true; return SCEVCheckBlock; } BasicBlock *InnerLoopVectorizer::emitMemRuntimeChecks(BasicBlock *Bypass) { // VPlan-native path does not do any analysis for runtime checks currently. if (EnableVPlanNativePath) return nullptr; BasicBlock *const MemCheckBlock = RTChecks.emitMemRuntimeChecks(Bypass, LoopVectorPreHeader); // Check if we generated code that checks in runtime if arrays overlap. We put // the checks into a separate block to make the more common case of few // elements faster. if (!MemCheckBlock) return nullptr; if (MemCheckBlock->getParent()->hasOptSize() || OptForSizeBasedOnProfile) { assert(Cost->Hints->getForce() == LoopVectorizeHints::FK_Enabled && "Cannot emit memory checks when optimizing for size, unless forced " "to vectorize."); ORE->emit([&]() { return OptimizationRemarkAnalysis(DEBUG_TYPE, "VectorizationCodeSize", OrigLoop->getStartLoc(), OrigLoop->getHeader()) << "Code-size may be reduced by not forcing " "vectorization, or by source-code modifications " "eliminating the need for runtime checks " "(e.g., adding 'restrict')."; }); } LoopBypassBlocks.push_back(MemCheckBlock); AddedSafetyChecks = true; return MemCheckBlock; } void InnerLoopVectorizer::createVectorLoopSkeleton(StringRef Prefix) { LoopScalarBody = OrigLoop->getHeader(); LoopVectorPreHeader = OrigLoop->getLoopPreheader(); assert(LoopVectorPreHeader && "Invalid loop structure"); LoopExitBlock = OrigLoop->getUniqueExitBlock(); // may be nullptr assert((LoopExitBlock || Cost->requiresScalarEpilogue(VF.isVector())) && "multiple exit loop without required epilogue?"); LoopMiddleBlock = SplitBlock(LoopVectorPreHeader, LoopVectorPreHeader->getTerminator(), DT, LI, nullptr, Twine(Prefix) + "middle.block"); LoopScalarPreHeader = SplitBlock(LoopMiddleBlock, LoopMiddleBlock->getTerminator(), DT, LI, nullptr, Twine(Prefix) + "scalar.ph"); auto *ScalarLatchTerm = OrigLoop->getLoopLatch()->getTerminator(); // Set up the middle block terminator. Two cases: // 1) If we know that we must execute the scalar epilogue, emit an // unconditional branch. // 2) Otherwise, we must have a single unique exit block (due to how we // implement the multiple exit case). In this case, set up a conditional // branch from the middle block to the loop scalar preheader, and the // exit block. completeLoopSkeleton will update the condition to use an // iteration check, if required to decide whether to execute the remainder. BranchInst *BrInst = Cost->requiresScalarEpilogue(VF.isVector()) ? BranchInst::Create(LoopScalarPreHeader) : BranchInst::Create(LoopExitBlock, LoopScalarPreHeader, Builder.getTrue()); BrInst->setDebugLoc(ScalarLatchTerm->getDebugLoc()); ReplaceInstWithInst(LoopMiddleBlock->getTerminator(), BrInst); // Update dominator for loop exit. During skeleton creation, only the vector // pre-header and the middle block are created. The vector loop is entirely // created during VPlan exection. if (!Cost->requiresScalarEpilogue(VF.isVector())) // If there is an epilogue which must run, there's no edge from the // middle block to exit blocks and thus no need to update the immediate // dominator of the exit blocks. DT->changeImmediateDominator(LoopExitBlock, LoopMiddleBlock); } PHINode *InnerLoopVectorizer::createInductionResumeValue( PHINode *OrigPhi, const InductionDescriptor &II, Value *Step, ArrayRef BypassBlocks, std::pair AdditionalBypass) { Value *VectorTripCount = getOrCreateVectorTripCount(LoopVectorPreHeader); assert(VectorTripCount && "Expected valid arguments"); Instruction *OldInduction = Legal->getPrimaryInduction(); Value *&EndValue = IVEndValues[OrigPhi]; Value *EndValueFromAdditionalBypass = AdditionalBypass.second; if (OrigPhi == OldInduction) { // We know what the end value is. EndValue = VectorTripCount; } else { IRBuilder<> B(LoopVectorPreHeader->getTerminator()); // Fast-math-flags propagate from the original induction instruction. if (II.getInductionBinOp() && isa(II.getInductionBinOp())) B.setFastMathFlags(II.getInductionBinOp()->getFastMathFlags()); EndValue = emitTransformedIndex(B, VectorTripCount, II.getStartValue(), Step, II.getKind(), II.getInductionBinOp()); EndValue->setName("ind.end"); // Compute the end value for the additional bypass (if applicable). if (AdditionalBypass.first) { B.SetInsertPoint(AdditionalBypass.first, AdditionalBypass.first->getFirstInsertionPt()); EndValueFromAdditionalBypass = emitTransformedIndex(B, AdditionalBypass.second, II.getStartValue(), Step, II.getKind(), II.getInductionBinOp()); EndValueFromAdditionalBypass->setName("ind.end"); } } // Create phi nodes to merge from the backedge-taken check block. PHINode *BCResumeVal = PHINode::Create(OrigPhi->getType(), 3, "bc.resume.val", LoopScalarPreHeader->getTerminator()); // Copy original phi DL over to the new one. BCResumeVal->setDebugLoc(OrigPhi->getDebugLoc()); // The new PHI merges the original incoming value, in case of a bypass, // or the value at the end of the vectorized loop. BCResumeVal->addIncoming(EndValue, LoopMiddleBlock); // Fix the scalar body counter (PHI node). // The old induction's phi node in the scalar body needs the truncated // value. for (BasicBlock *BB : BypassBlocks) BCResumeVal->addIncoming(II.getStartValue(), BB); if (AdditionalBypass.first) BCResumeVal->setIncomingValueForBlock(AdditionalBypass.first, EndValueFromAdditionalBypass); return BCResumeVal; } /// Return the expanded step for \p ID using \p ExpandedSCEVs to look up SCEV /// expansion results. static Value *getExpandedStep(const InductionDescriptor &ID, const SCEV2ValueTy &ExpandedSCEVs) { const SCEV *Step = ID.getStep(); if (auto *C = dyn_cast(Step)) return C->getValue(); if (auto *U = dyn_cast(Step)) return U->getValue(); auto I = ExpandedSCEVs.find(Step); assert(I != ExpandedSCEVs.end() && "SCEV must be expanded at this point"); return I->second; } void InnerLoopVectorizer::createInductionResumeValues( const SCEV2ValueTy &ExpandedSCEVs, std::pair AdditionalBypass) { assert(((AdditionalBypass.first && AdditionalBypass.second) || (!AdditionalBypass.first && !AdditionalBypass.second)) && "Inconsistent information about additional bypass."); // We are going to resume the execution of the scalar loop. // Go over all of the induction variables that we found and fix the // PHIs that are left in the scalar version of the loop. // The starting values of PHI nodes depend on the counter of the last // iteration in the vectorized loop. // If we come from a bypass edge then we need to start from the original // start value. for (const auto &InductionEntry : Legal->getInductionVars()) { PHINode *OrigPhi = InductionEntry.first; const InductionDescriptor &II = InductionEntry.second; PHINode *BCResumeVal = createInductionResumeValue( OrigPhi, II, getExpandedStep(II, ExpandedSCEVs), LoopBypassBlocks, AdditionalBypass); OrigPhi->setIncomingValueForBlock(LoopScalarPreHeader, BCResumeVal); } } BasicBlock *InnerLoopVectorizer::completeLoopSkeleton() { // The trip counts should be cached by now. Value *Count = getTripCount(); Value *VectorTripCount = getOrCreateVectorTripCount(LoopVectorPreHeader); auto *ScalarLatchTerm = OrigLoop->getLoopLatch()->getTerminator(); // Add a check in the middle block to see if we have completed // all of the iterations in the first vector loop. Three cases: // 1) If we require a scalar epilogue, there is no conditional branch as // we unconditionally branch to the scalar preheader. Do nothing. // 2) If (N - N%VF) == N, then we *don't* need to run the remainder. // Thus if tail is to be folded, we know we don't need to run the // remainder and we can use the previous value for the condition (true). // 3) Otherwise, construct a runtime check. if (!Cost->requiresScalarEpilogue(VF.isVector()) && !Cost->foldTailByMasking()) { // Here we use the same DebugLoc as the scalar loop latch terminator instead // of the corresponding compare because they may have ended up with // different line numbers and we want to avoid awkward line stepping while // debugging. Eg. if the compare has got a line number inside the loop. // TODO: At the moment, CreateICmpEQ will simplify conditions with constant // operands. Perform simplification directly on VPlan once the branch is // modeled there. IRBuilder<> B(LoopMiddleBlock->getTerminator()); B.SetCurrentDebugLocation(ScalarLatchTerm->getDebugLoc()); Value *CmpN = B.CreateICmpEQ(Count, VectorTripCount, "cmp.n"); BranchInst &BI = *cast(LoopMiddleBlock->getTerminator()); BI.setCondition(CmpN); if (hasBranchWeightMD(*ScalarLatchTerm)) { // Assume that `Count % VectorTripCount` is equally distributed. unsigned TripCount = UF * VF.getKnownMinValue(); assert(TripCount > 0 && "trip count should not be zero"); const uint32_t Weights[] = {1, TripCount - 1}; setBranchWeights(BI, Weights); } } #ifdef EXPENSIVE_CHECKS assert(DT->verify(DominatorTree::VerificationLevel::Fast)); #endif return LoopVectorPreHeader; } std::pair InnerLoopVectorizer::createVectorizedLoopSkeleton( const SCEV2ValueTy &ExpandedSCEVs) { /* In this function we generate a new loop. The new loop will contain the vectorized instructions while the old loop will continue to run the scalar remainder. [ ] <-- old preheader - loop iteration number check and SCEVs in Plan's / | preheader are expanded here. Eventually all required SCEV / | expansion should happen here. / v | [ ] <-- vector loop bypass (may consist of multiple blocks). | / | | / v || [ ] <-- vector pre header. |/ | | v | [ ] \ | [ ]_| <-- vector loop (created during VPlan execution). | | | v \ -[ ] <--- middle-block. \/ | /\ v | ->[ ] <--- new preheader. | | (opt) v <-- edge from middle to exit iff epilogue is not required. | [ ] \ | [ ]_| <-- old scalar loop to handle remainder (scalar epilogue). \ | \ v >[ ] <-- exit block(s). ... */ // Create an empty vector loop, and prepare basic blocks for the runtime // checks. createVectorLoopSkeleton(""); // Now, compare the new count to zero. If it is zero skip the vector loop and // jump to the scalar loop. This check also covers the case where the // backedge-taken count is uint##_max: adding one to it will overflow leading // to an incorrect trip count of zero. In this (rare) case we will also jump // to the scalar loop. emitIterationCountCheck(LoopScalarPreHeader); // Generate the code to check any assumptions that we've made for SCEV // expressions. emitSCEVChecks(LoopScalarPreHeader); // Generate the code that checks in runtime if arrays overlap. We put the // checks into a separate block to make the more common case of few elements // faster. emitMemRuntimeChecks(LoopScalarPreHeader); // Emit phis for the new starting index of the scalar loop. createInductionResumeValues(ExpandedSCEVs); return {completeLoopSkeleton(), nullptr}; } // Fix up external users of the induction variable. At this point, we are // in LCSSA form, with all external PHIs that use the IV having one input value, // coming from the remainder loop. We need those PHIs to also have a correct // value for the IV when arriving directly from the middle block. void InnerLoopVectorizer::fixupIVUsers(PHINode *OrigPhi, const InductionDescriptor &II, Value *VectorTripCount, Value *EndValue, BasicBlock *MiddleBlock, BasicBlock *VectorHeader, VPlan &Plan, VPTransformState &State) { // There are two kinds of external IV usages - those that use the value // computed in the last iteration (the PHI) and those that use the penultimate // value (the value that feeds into the phi from the loop latch). // We allow both, but they, obviously, have different values. assert(OrigLoop->getUniqueExitBlock() && "Expected a single exit block"); DenseMap MissingVals; // An external user of the last iteration's value should see the value that // the remainder loop uses to initialize its own IV. Value *PostInc = OrigPhi->getIncomingValueForBlock(OrigLoop->getLoopLatch()); for (User *U : PostInc->users()) { Instruction *UI = cast(U); if (!OrigLoop->contains(UI)) { assert(isa(UI) && "Expected LCSSA form"); MissingVals[UI] = EndValue; } } // An external user of the penultimate value need to see EndValue - Step. // The simplest way to get this is to recompute it from the constituent SCEVs, // that is Start + (Step * (CRD - 1)). for (User *U : OrigPhi->users()) { auto *UI = cast(U); if (!OrigLoop->contains(UI)) { assert(isa(UI) && "Expected LCSSA form"); IRBuilder<> B(MiddleBlock->getTerminator()); // Fast-math-flags propagate from the original induction instruction. if (II.getInductionBinOp() && isa(II.getInductionBinOp())) B.setFastMathFlags(II.getInductionBinOp()->getFastMathFlags()); Value *CountMinusOne = B.CreateSub( VectorTripCount, ConstantInt::get(VectorTripCount->getType(), 1)); CountMinusOne->setName("cmo"); VPValue *StepVPV = Plan.getSCEVExpansion(II.getStep()); assert(StepVPV && "step must have been expanded during VPlan execution"); Value *Step = StepVPV->isLiveIn() ? StepVPV->getLiveInIRValue() : State.get(StepVPV, {0, 0}); Value *Escape = emitTransformedIndex(B, CountMinusOne, II.getStartValue(), Step, II.getKind(), II.getInductionBinOp()); Escape->setName("ind.escape"); MissingVals[UI] = Escape; } } for (auto &I : MissingVals) { PHINode *PHI = cast(I.first); // One corner case we have to handle is two IVs "chasing" each-other, // that is %IV2 = phi [...], [ %IV1, %latch ] // In this case, if IV1 has an external use, we need to avoid adding both // "last value of IV1" and "penultimate value of IV2". So, verify that we // don't already have an incoming value for the middle block. if (PHI->getBasicBlockIndex(MiddleBlock) == -1) { PHI->addIncoming(I.second, MiddleBlock); Plan.removeLiveOut(PHI); } } } namespace { struct CSEDenseMapInfo { static bool canHandle(const Instruction *I) { return isa(I) || isa(I) || isa(I) || isa(I); } static inline Instruction *getEmptyKey() { return DenseMapInfo::getEmptyKey(); } static inline Instruction *getTombstoneKey() { return DenseMapInfo::getTombstoneKey(); } static unsigned getHashValue(const Instruction *I) { assert(canHandle(I) && "Unknown instruction!"); return hash_combine(I->getOpcode(), hash_combine_range(I->value_op_begin(), I->value_op_end())); } static bool isEqual(const Instruction *LHS, const Instruction *RHS) { if (LHS == getEmptyKey() || RHS == getEmptyKey() || LHS == getTombstoneKey() || RHS == getTombstoneKey()) return LHS == RHS; return LHS->isIdenticalTo(RHS); } }; } // end anonymous namespace ///Perform cse of induction variable instructions. static void cse(BasicBlock *BB) { // Perform simple cse. SmallDenseMap CSEMap; for (Instruction &In : llvm::make_early_inc_range(*BB)) { if (!CSEDenseMapInfo::canHandle(&In)) continue; // Check if we can replace this instruction with any of the // visited instructions. if (Instruction *V = CSEMap.lookup(&In)) { In.replaceAllUsesWith(V); In.eraseFromParent(); continue; } CSEMap[&In] = &In; } } InstructionCost LoopVectorizationCostModel::getVectorCallCost(CallInst *CI, ElementCount VF) const { // We only need to calculate a cost if the VF is scalar; for actual vectors // we should already have a pre-calculated cost at each VF. if (!VF.isScalar()) return CallWideningDecisions.at(std::make_pair(CI, VF)).Cost; TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput; Type *RetTy = CI->getType(); if (RecurrenceDescriptor::isFMulAddIntrinsic(CI)) if (auto RedCost = getReductionPatternCost(CI, VF, RetTy, CostKind)) return *RedCost; SmallVector Tys; for (auto &ArgOp : CI->args()) Tys.push_back(ArgOp->getType()); InstructionCost ScalarCallCost = TTI.getCallInstrCost(CI->getCalledFunction(), RetTy, Tys, CostKind); // If this is an intrinsic we may have a lower cost for it. if (getVectorIntrinsicIDForCall(CI, TLI)) { InstructionCost IntrinsicCost = getVectorIntrinsicCost(CI, VF); return std::min(ScalarCallCost, IntrinsicCost); } return ScalarCallCost; } static Type *MaybeVectorizeType(Type *Elt, ElementCount VF) { if (VF.isScalar() || (!Elt->isIntOrPtrTy() && !Elt->isFloatingPointTy())) return Elt; return VectorType::get(Elt, VF); } InstructionCost LoopVectorizationCostModel::getVectorIntrinsicCost(CallInst *CI, ElementCount VF) const { Intrinsic::ID ID = getVectorIntrinsicIDForCall(CI, TLI); assert(ID && "Expected intrinsic call!"); Type *RetTy = MaybeVectorizeType(CI->getType(), VF); FastMathFlags FMF; if (auto *FPMO = dyn_cast(CI)) FMF = FPMO->getFastMathFlags(); SmallVector Arguments(CI->args()); FunctionType *FTy = CI->getCalledFunction()->getFunctionType(); SmallVector ParamTys; std::transform(FTy->param_begin(), FTy->param_end(), std::back_inserter(ParamTys), [&](Type *Ty) { return MaybeVectorizeType(Ty, VF); }); IntrinsicCostAttributes CostAttrs(ID, RetTy, Arguments, ParamTys, FMF, dyn_cast(CI)); return TTI.getIntrinsicInstrCost(CostAttrs, TargetTransformInfo::TCK_RecipThroughput); } static Type *smallestIntegerVectorType(Type *T1, Type *T2) { auto *I1 = cast(cast(T1)->getElementType()); auto *I2 = cast(cast(T2)->getElementType()); return I1->getBitWidth() < I2->getBitWidth() ? T1 : T2; } static Type *largestIntegerVectorType(Type *T1, Type *T2) { auto *I1 = cast(cast(T1)->getElementType()); auto *I2 = cast(cast(T2)->getElementType()); return I1->getBitWidth() > I2->getBitWidth() ? T1 : T2; } void InnerLoopVectorizer::fixVectorizedLoop(VPTransformState &State, VPlan &Plan) { // Fix widened non-induction PHIs by setting up the PHI operands. if (EnableVPlanNativePath) fixNonInductionPHIs(Plan, State); // At this point every instruction in the original loop is widened to a // vector form. Now we need to fix the recurrences in the loop. These PHI // nodes are currently empty because we did not want to introduce cycles. // This is the second stage of vectorizing recurrences. Note that fixing // reduction phis are already modeled in VPlan. // TODO: Also model fixing fixed-order recurrence phis in VPlan. VPRegionBlock *VectorRegion = State.Plan->getVectorLoopRegion(); VPBasicBlock *HeaderVPBB = VectorRegion->getEntryBasicBlock(); for (VPRecipeBase &R : HeaderVPBB->phis()) { if (auto *FOR = dyn_cast(&R)) fixFixedOrderRecurrence(FOR, State); } // Forget the original basic block. PSE.getSE()->forgetLoop(OrigLoop); PSE.getSE()->forgetBlockAndLoopDispositions(); // After vectorization, the exit blocks of the original loop will have // additional predecessors. Invalidate SCEVs for the exit phis in case SE // looked through single-entry phis. SmallVector ExitBlocks; OrigLoop->getExitBlocks(ExitBlocks); for (BasicBlock *Exit : ExitBlocks) for (PHINode &PN : Exit->phis()) PSE.getSE()->forgetLcssaPhiWithNewPredecessor(OrigLoop, &PN); VPBasicBlock *LatchVPBB = VectorRegion->getExitingBasicBlock(); Loop *VectorLoop = LI->getLoopFor(State.CFG.VPBB2IRBB[LatchVPBB]); if (Cost->requiresScalarEpilogue(VF.isVector())) { // No edge from the middle block to the unique exit block has been inserted // and there is nothing to fix from vector loop; phis should have incoming // from scalar loop only. } else { // TODO: Check VPLiveOuts to see if IV users need fixing instead of checking // the cost model. // If we inserted an edge from the middle block to the unique exit block, // update uses outside the loop (phis) to account for the newly inserted // edge. // Fix-up external users of the induction variables. for (const auto &Entry : Legal->getInductionVars()) fixupIVUsers(Entry.first, Entry.second, getOrCreateVectorTripCount(VectorLoop->getLoopPreheader()), IVEndValues[Entry.first], LoopMiddleBlock, VectorLoop->getHeader(), Plan, State); } // Fix LCSSA phis not already fixed earlier. Extracts may need to be generated // in the exit block, so update the builder. State.Builder.SetInsertPoint(State.CFG.ExitBB, State.CFG.ExitBB->getFirstNonPHIIt()); for (const auto &KV : Plan.getLiveOuts()) KV.second->fixPhi(Plan, State); for (Instruction *PI : PredicatedInstructions) sinkScalarOperands(&*PI); // Remove redundant induction instructions. cse(VectorLoop->getHeader()); // Set/update profile weights for the vector and remainder loops as original // loop iterations are now distributed among them. Note that original loop // represented by LoopScalarBody becomes remainder loop after vectorization. // // For cases like foldTailByMasking() and requiresScalarEpiloque() we may // end up getting slightly roughened result but that should be OK since // profile is not inherently precise anyway. Note also possible bypass of // vector code caused by legality checks is ignored, assigning all the weight // to the vector loop, optimistically. // // For scalable vectorization we can't know at compile time how many iterations // of the loop are handled in one vector iteration, so instead assume a pessimistic // vscale of '1'. setProfileInfoAfterUnrolling(LI->getLoopFor(LoopScalarBody), VectorLoop, LI->getLoopFor(LoopScalarBody), VF.getKnownMinValue() * UF); } void InnerLoopVectorizer::fixFixedOrderRecurrence( VPFirstOrderRecurrencePHIRecipe *PhiR, VPTransformState &State) { // This is the second phase of vectorizing first-order recurrences. An // overview of the transformation is described below. Suppose we have the // following loop. // // for (int i = 0; i < n; ++i) // b[i] = a[i] - a[i - 1]; // // There is a first-order recurrence on "a". For this loop, the shorthand // scalar IR looks like: // // scalar.ph: // s_init = a[-1] // br scalar.body // // scalar.body: // i = phi [0, scalar.ph], [i+1, scalar.body] // s1 = phi [s_init, scalar.ph], [s2, scalar.body] // s2 = a[i] // b[i] = s2 - s1 // br cond, scalar.body, ... // // In this example, s1 is a recurrence because it's value depends on the // previous iteration. In the first phase of vectorization, we created a // vector phi v1 for s1. We now complete the vectorization and produce the // shorthand vector IR shown below (for VF = 4, UF = 1). // // vector.ph: // v_init = vector(..., ..., ..., a[-1]) // br vector.body // // vector.body // i = phi [0, vector.ph], [i+4, vector.body] // v1 = phi [v_init, vector.ph], [v2, vector.body] // v2 = a[i, i+1, i+2, i+3]; // v3 = vector(v1(3), v2(0, 1, 2)) // b[i, i+1, i+2, i+3] = v2 - v3 // br cond, vector.body, middle.block // // middle.block: // x = v2(3) // br scalar.ph // // scalar.ph: // s_init = phi [x, middle.block], [a[-1], otherwise] // br scalar.body // // After execution completes the vector loop, we extract the next value of // the recurrence (x) to use as the initial value in the scalar loop. // Extract the last vector element in the middle block. This will be the // initial value for the recurrence when jumping to the scalar loop. VPValue *PreviousDef = PhiR->getBackedgeValue(); Value *Incoming = State.get(PreviousDef, UF - 1); auto *ExtractForScalar = Incoming; auto *IdxTy = Builder.getInt32Ty(); Value *RuntimeVF = nullptr; if (VF.isVector()) { auto *One = ConstantInt::get(IdxTy, 1); Builder.SetInsertPoint(LoopMiddleBlock->getTerminator()); RuntimeVF = getRuntimeVF(Builder, IdxTy, VF); auto *LastIdx = Builder.CreateSub(RuntimeVF, One); ExtractForScalar = Builder.CreateExtractElement(Incoming, LastIdx, "vector.recur.extract"); } auto RecurSplice = cast(*PhiR->user_begin()); assert(PhiR->getNumUsers() == 1 && RecurSplice->getOpcode() == VPInstruction::FirstOrderRecurrenceSplice && "recurrence phi must have a single user: FirstOrderRecurrenceSplice"); SmallVector LiveOuts; for (VPUser *U : RecurSplice->users()) if (auto *LiveOut = dyn_cast(U)) LiveOuts.push_back(LiveOut); if (!LiveOuts.empty()) { // Extract the second last element in the middle block if the // Phi is used outside the loop. We need to extract the phi itself // and not the last element (the phi update in the current iteration). This // will be the value when jumping to the exit block from the // LoopMiddleBlock, when the scalar loop is not run at all. Value *ExtractForPhiUsedOutsideLoop = nullptr; if (VF.isVector()) { auto *Idx = Builder.CreateSub(RuntimeVF, ConstantInt::get(IdxTy, 2)); ExtractForPhiUsedOutsideLoop = Builder.CreateExtractElement( Incoming, Idx, "vector.recur.extract.for.phi"); } else { assert(UF > 1 && "VF and UF cannot both be 1"); // When loop is unrolled without vectorizing, initialize // ExtractForPhiUsedOutsideLoop with the value just prior to unrolled // value of `Incoming`. This is analogous to the vectorized case above: // extracting the second last element when VF > 1. ExtractForPhiUsedOutsideLoop = State.get(PreviousDef, UF - 2); } for (VPLiveOut *LiveOut : LiveOuts) { assert(!Cost->requiresScalarEpilogue(VF.isVector())); PHINode *LCSSAPhi = LiveOut->getPhi(); LCSSAPhi->addIncoming(ExtractForPhiUsedOutsideLoop, LoopMiddleBlock); State.Plan->removeLiveOut(LCSSAPhi); } } // Fix the initial value of the original recurrence in the scalar loop. Builder.SetInsertPoint(LoopScalarPreHeader, LoopScalarPreHeader->begin()); PHINode *Phi = cast(PhiR->getUnderlyingValue()); auto *Start = Builder.CreatePHI(Phi->getType(), 2, "scalar.recur.init"); auto *ScalarInit = PhiR->getStartValue()->getLiveInIRValue(); for (auto *BB : predecessors(LoopScalarPreHeader)) { auto *Incoming = BB == LoopMiddleBlock ? ExtractForScalar : ScalarInit; Start->addIncoming(Incoming, BB); } Phi->setIncomingValueForBlock(LoopScalarPreHeader, Start); Phi->setName("scalar.recur"); } void InnerLoopVectorizer::sinkScalarOperands(Instruction *PredInst) { // The basic block and loop containing the predicated instruction. auto *PredBB = PredInst->getParent(); auto *VectorLoop = LI->getLoopFor(PredBB); // Initialize a worklist with the operands of the predicated instruction. SetVector Worklist(PredInst->op_begin(), PredInst->op_end()); // Holds instructions that we need to analyze again. An instruction may be // reanalyzed if we don't yet know if we can sink it or not. SmallVector InstsToReanalyze; // Returns true if a given use occurs in the predicated block. Phi nodes use // their operands in their corresponding predecessor blocks. auto isBlockOfUsePredicated = [&](Use &U) -> bool { auto *I = cast(U.getUser()); BasicBlock *BB = I->getParent(); if (auto *Phi = dyn_cast(I)) BB = Phi->getIncomingBlock( PHINode::getIncomingValueNumForOperand(U.getOperandNo())); return BB == PredBB; }; // Iteratively sink the scalarized operands of the predicated instruction // into the block we created for it. When an instruction is sunk, it's // operands are then added to the worklist. The algorithm ends after one pass // through the worklist doesn't sink a single instruction. bool Changed; do { // Add the instructions that need to be reanalyzed to the worklist, and // reset the changed indicator. Worklist.insert(InstsToReanalyze.begin(), InstsToReanalyze.end()); InstsToReanalyze.clear(); Changed = false; while (!Worklist.empty()) { auto *I = dyn_cast(Worklist.pop_back_val()); // We can't sink an instruction if it is a phi node, is not in the loop, // may have side effects or may read from memory. // TODO Could dor more granular checking to allow sinking a load past non-store instructions. if (!I || isa(I) || !VectorLoop->contains(I) || I->mayHaveSideEffects() || I->mayReadFromMemory()) continue; // If the instruction is already in PredBB, check if we can sink its // operands. In that case, VPlan's sinkScalarOperands() succeeded in // sinking the scalar instruction I, hence it appears in PredBB; but it // may have failed to sink I's operands (recursively), which we try // (again) here. if (I->getParent() == PredBB) { Worklist.insert(I->op_begin(), I->op_end()); continue; } // It's legal to sink the instruction if all its uses occur in the // predicated block. Otherwise, there's nothing to do yet, and we may // need to reanalyze the instruction. if (!llvm::all_of(I->uses(), isBlockOfUsePredicated)) { InstsToReanalyze.push_back(I); continue; } // Move the instruction to the beginning of the predicated block, and add // it's operands to the worklist. I->moveBefore(&*PredBB->getFirstInsertionPt()); Worklist.insert(I->op_begin(), I->op_end()); // The sinking may have enabled other instructions to be sunk, so we will // need to iterate. Changed = true; } } while (Changed); } void InnerLoopVectorizer::fixNonInductionPHIs(VPlan &Plan, VPTransformState &State) { auto Iter = vp_depth_first_deep(Plan.getEntry()); for (VPBasicBlock *VPBB : VPBlockUtils::blocksOnly(Iter)) { for (VPRecipeBase &P : VPBB->phis()) { VPWidenPHIRecipe *VPPhi = dyn_cast(&P); if (!VPPhi) continue; PHINode *NewPhi = cast(State.get(VPPhi, 0)); // Make sure the builder has a valid insert point. Builder.SetInsertPoint(NewPhi); for (unsigned i = 0; i < VPPhi->getNumOperands(); ++i) { VPValue *Inc = VPPhi->getIncomingValue(i); VPBasicBlock *VPBB = VPPhi->getIncomingBlock(i); NewPhi->addIncoming(State.get(Inc, 0), State.CFG.VPBB2IRBB[VPBB]); } } } } bool InnerLoopVectorizer::useOrderedReductions( const RecurrenceDescriptor &RdxDesc) { return Cost->useOrderedReductions(RdxDesc); } void LoopVectorizationCostModel::collectLoopScalars(ElementCount VF) { // We should not collect Scalars more than once per VF. Right now, this // function is called from collectUniformsAndScalars(), which already does // this check. Collecting Scalars for VF=1 does not make any sense. assert(VF.isVector() && !Scalars.contains(VF) && "This function should not be visited twice for the same VF"); // This avoids any chances of creating a REPLICATE recipe during planning // since that would result in generation of scalarized code during execution, // which is not supported for scalable vectors. if (VF.isScalable()) { Scalars[VF].insert(Uniforms[VF].begin(), Uniforms[VF].end()); return; } SmallSetVector Worklist; // These sets are used to seed the analysis with pointers used by memory // accesses that will remain scalar. SmallSetVector ScalarPtrs; SmallPtrSet PossibleNonScalarPtrs; auto *Latch = TheLoop->getLoopLatch(); // A helper that returns true if the use of Ptr by MemAccess will be scalar. // The pointer operands of loads and stores will be scalar as long as the // memory access is not a gather or scatter operation. The value operand of a // store will remain scalar if the store is scalarized. auto isScalarUse = [&](Instruction *MemAccess, Value *Ptr) { InstWidening WideningDecision = getWideningDecision(MemAccess, VF); assert(WideningDecision != CM_Unknown && "Widening decision should be ready at this moment"); if (auto *Store = dyn_cast(MemAccess)) if (Ptr == Store->getValueOperand()) return WideningDecision == CM_Scalarize; assert(Ptr == getLoadStorePointerOperand(MemAccess) && "Ptr is neither a value or pointer operand"); return WideningDecision != CM_GatherScatter; }; // A helper that returns true if the given value is a bitcast or // getelementptr instruction contained in the loop. auto isLoopVaryingBitCastOrGEP = [&](Value *V) { return ((isa(V) && V->getType()->isPointerTy()) || isa(V)) && !TheLoop->isLoopInvariant(V); }; // A helper that evaluates a memory access's use of a pointer. If the use will // be a scalar use and the pointer is only used by memory accesses, we place // the pointer in ScalarPtrs. Otherwise, the pointer is placed in // PossibleNonScalarPtrs. auto evaluatePtrUse = [&](Instruction *MemAccess, Value *Ptr) { // We only care about bitcast and getelementptr instructions contained in // the loop. if (!isLoopVaryingBitCastOrGEP(Ptr)) return; // If the pointer has already been identified as scalar (e.g., if it was // also identified as uniform), there's nothing to do. auto *I = cast(Ptr); if (Worklist.count(I)) return; // If the use of the pointer will be a scalar use, and all users of the // pointer are memory accesses, place the pointer in ScalarPtrs. Otherwise, // place the pointer in PossibleNonScalarPtrs. if (isScalarUse(MemAccess, Ptr) && llvm::all_of(I->users(), [&](User *U) { return isa(U) || isa(U); })) ScalarPtrs.insert(I); else PossibleNonScalarPtrs.insert(I); }; // We seed the scalars analysis with three classes of instructions: (1) // instructions marked uniform-after-vectorization and (2) bitcast, // getelementptr and (pointer) phi instructions used by memory accesses // requiring a scalar use. // // (1) Add to the worklist all instructions that have been identified as // uniform-after-vectorization. Worklist.insert(Uniforms[VF].begin(), Uniforms[VF].end()); // (2) Add to the worklist all bitcast and getelementptr instructions used by // memory accesses requiring a scalar use. The pointer operands of loads and // stores will be scalar as long as the memory accesses is not a gather or // scatter operation. The value operand of a store will remain scalar if the // store is scalarized. for (auto *BB : TheLoop->blocks()) for (auto &I : *BB) { if (auto *Load = dyn_cast(&I)) { evaluatePtrUse(Load, Load->getPointerOperand()); } else if (auto *Store = dyn_cast(&I)) { evaluatePtrUse(Store, Store->getPointerOperand()); evaluatePtrUse(Store, Store->getValueOperand()); } } for (auto *I : ScalarPtrs) if (!PossibleNonScalarPtrs.count(I)) { LLVM_DEBUG(dbgs() << "LV: Found scalar instruction: " << *I << "\n"); Worklist.insert(I); } // Insert the forced scalars. // FIXME: Currently VPWidenPHIRecipe() often creates a dead vector // induction variable when the PHI user is scalarized. auto ForcedScalar = ForcedScalars.find(VF); if (ForcedScalar != ForcedScalars.end()) for (auto *I : ForcedScalar->second) { LLVM_DEBUG(dbgs() << "LV: Found (forced) scalar instruction: " << *I << "\n"); Worklist.insert(I); } // Expand the worklist by looking through any bitcasts and getelementptr // instructions we've already identified as scalar. This is similar to the // expansion step in collectLoopUniforms(); however, here we're only // expanding to include additional bitcasts and getelementptr instructions. unsigned Idx = 0; while (Idx != Worklist.size()) { Instruction *Dst = Worklist[Idx++]; if (!isLoopVaryingBitCastOrGEP(Dst->getOperand(0))) continue; auto *Src = cast(Dst->getOperand(0)); if (llvm::all_of(Src->users(), [&](User *U) -> bool { auto *J = cast(U); return !TheLoop->contains(J) || Worklist.count(J) || ((isa(J) || isa(J)) && isScalarUse(J, Src)); })) { Worklist.insert(Src); LLVM_DEBUG(dbgs() << "LV: Found scalar instruction: " << *Src << "\n"); } } // An induction variable will remain scalar if all users of the induction // variable and induction variable update remain scalar. for (const auto &Induction : Legal->getInductionVars()) { auto *Ind = Induction.first; auto *IndUpdate = cast(Ind->getIncomingValueForBlock(Latch)); // If tail-folding is applied, the primary induction variable will be used // to feed a vector compare. if (Ind == Legal->getPrimaryInduction() && foldTailByMasking()) continue; // Returns true if \p Indvar is a pointer induction that is used directly by // load/store instruction \p I. auto IsDirectLoadStoreFromPtrIndvar = [&](Instruction *Indvar, Instruction *I) { return Induction.second.getKind() == InductionDescriptor::IK_PtrInduction && (isa(I) || isa(I)) && Indvar == getLoadStorePointerOperand(I) && isScalarUse(I, Indvar); }; // Determine if all users of the induction variable are scalar after // vectorization. auto ScalarInd = llvm::all_of(Ind->users(), [&](User *U) -> bool { auto *I = cast(U); return I == IndUpdate || !TheLoop->contains(I) || Worklist.count(I) || IsDirectLoadStoreFromPtrIndvar(Ind, I); }); if (!ScalarInd) continue; // Determine if all users of the induction variable update instruction are // scalar after vectorization. auto ScalarIndUpdate = llvm::all_of(IndUpdate->users(), [&](User *U) -> bool { auto *I = cast(U); return I == Ind || !TheLoop->contains(I) || Worklist.count(I) || IsDirectLoadStoreFromPtrIndvar(IndUpdate, I); }); if (!ScalarIndUpdate) continue; // The induction variable and its update instruction will remain scalar. Worklist.insert(Ind); Worklist.insert(IndUpdate); LLVM_DEBUG(dbgs() << "LV: Found scalar instruction: " << *Ind << "\n"); LLVM_DEBUG(dbgs() << "LV: Found scalar instruction: " << *IndUpdate << "\n"); } Scalars[VF].insert(Worklist.begin(), Worklist.end()); } bool LoopVectorizationCostModel::isScalarWithPredication( Instruction *I, ElementCount VF) const { if (!isPredicatedInst(I)) return false; // Do we have a non-scalar lowering for this predicated // instruction? No - it is scalar with predication. switch(I->getOpcode()) { default: return true; case Instruction::Call: if (VF.isScalar()) return true; return CallWideningDecisions.at(std::make_pair(cast(I), VF)) .Kind == CM_Scalarize; case Instruction::Load: case Instruction::Store: { auto *Ptr = getLoadStorePointerOperand(I); auto *Ty = getLoadStoreType(I); Type *VTy = Ty; if (VF.isVector()) VTy = VectorType::get(Ty, VF); const Align Alignment = getLoadStoreAlignment(I); return isa(I) ? !(isLegalMaskedLoad(Ty, Ptr, Alignment) || TTI.isLegalMaskedGather(VTy, Alignment)) : !(isLegalMaskedStore(Ty, Ptr, Alignment) || TTI.isLegalMaskedScatter(VTy, Alignment)); } case Instruction::UDiv: case Instruction::SDiv: case Instruction::SRem: case Instruction::URem: { // We have the option to use the safe-divisor idiom to avoid predication. // The cost based decision here will always select safe-divisor for // scalable vectors as scalarization isn't legal. const auto [ScalarCost, SafeDivisorCost] = getDivRemSpeculationCost(I, VF); return isDivRemScalarWithPredication(ScalarCost, SafeDivisorCost); } } } bool LoopVectorizationCostModel::isPredicatedInst(Instruction *I) const { if (!blockNeedsPredicationForAnyReason(I->getParent())) return false; // Can we prove this instruction is safe to unconditionally execute? // If not, we must use some form of predication. switch(I->getOpcode()) { default: return false; case Instruction::Load: case Instruction::Store: { if (!Legal->isMaskRequired(I)) return false; // When we know the load's address is loop invariant and the instruction // in the original scalar loop was unconditionally executed then we // don't need to mark it as a predicated instruction. Tail folding may // introduce additional predication, but we're guaranteed to always have // at least one active lane. We call Legal->blockNeedsPredication here // because it doesn't query tail-folding. For stores, we need to prove // both speculation safety (which follows from the same argument as loads), // but also must prove the value being stored is correct. The easiest // form of the later is to require that all values stored are the same. if (Legal->isInvariant(getLoadStorePointerOperand(I)) && (isa(I) || (isa(I) && TheLoop->isLoopInvariant(cast(I)->getValueOperand()))) && !Legal->blockNeedsPredication(I->getParent())) return false; return true; } case Instruction::UDiv: case Instruction::SDiv: case Instruction::SRem: case Instruction::URem: // TODO: We can use the loop-preheader as context point here and get // context sensitive reasoning return !isSafeToSpeculativelyExecute(I); case Instruction::Call: return Legal->isMaskRequired(I); } } std::pair LoopVectorizationCostModel::getDivRemSpeculationCost(Instruction *I, ElementCount VF) const { assert(I->getOpcode() == Instruction::UDiv || I->getOpcode() == Instruction::SDiv || I->getOpcode() == Instruction::SRem || I->getOpcode() == Instruction::URem); assert(!isSafeToSpeculativelyExecute(I)); const TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput; // Scalarization isn't legal for scalable vector types InstructionCost ScalarizationCost = InstructionCost::getInvalid(); if (!VF.isScalable()) { // Get the scalarization cost and scale this amount by the probability of // executing the predicated block. If the instruction is not predicated, // we fall through to the next case. ScalarizationCost = 0; // These instructions have a non-void type, so account for the phi nodes // that we will create. This cost is likely to be zero. The phi node // cost, if any, should be scaled by the block probability because it // models a copy at the end of each predicated block. ScalarizationCost += VF.getKnownMinValue() * TTI.getCFInstrCost(Instruction::PHI, CostKind); // The cost of the non-predicated instruction. ScalarizationCost += VF.getKnownMinValue() * TTI.getArithmeticInstrCost(I->getOpcode(), I->getType(), CostKind); // The cost of insertelement and extractelement instructions needed for // scalarization. ScalarizationCost += getScalarizationOverhead(I, VF, CostKind); // Scale the cost by the probability of executing the predicated blocks. // This assumes the predicated block for each vector lane is equally // likely. ScalarizationCost = ScalarizationCost / getReciprocalPredBlockProb(); } InstructionCost SafeDivisorCost = 0; auto *VecTy = ToVectorTy(I->getType(), VF); // The cost of the select guard to ensure all lanes are well defined // after we speculate above any internal control flow. SafeDivisorCost += TTI.getCmpSelInstrCost( Instruction::Select, VecTy, ToVectorTy(Type::getInt1Ty(I->getContext()), VF), CmpInst::BAD_ICMP_PREDICATE, CostKind); // Certain instructions can be cheaper to vectorize if they have a constant // second vector operand. One example of this are shifts on x86. Value *Op2 = I->getOperand(1); auto Op2Info = TTI.getOperandInfo(Op2); if (Op2Info.Kind == TargetTransformInfo::OK_AnyValue && Legal->isInvariant(Op2)) Op2Info.Kind = TargetTransformInfo::OK_UniformValue; SmallVector Operands(I->operand_values()); SafeDivisorCost += TTI.getArithmeticInstrCost( I->getOpcode(), VecTy, CostKind, {TargetTransformInfo::OK_AnyValue, TargetTransformInfo::OP_None}, Op2Info, Operands, I); return {ScalarizationCost, SafeDivisorCost}; } bool LoopVectorizationCostModel::interleavedAccessCanBeWidened( Instruction *I, ElementCount VF) { assert(isAccessInterleaved(I) && "Expecting interleaved access."); assert(getWideningDecision(I, VF) == CM_Unknown && "Decision should not be set yet."); auto *Group = getInterleavedAccessGroup(I); assert(Group && "Must have a group."); // If the instruction's allocated size doesn't equal it's type size, it // requires padding and will be scalarized. auto &DL = I->getModule()->getDataLayout(); auto *ScalarTy = getLoadStoreType(I); if (hasIrregularType(ScalarTy, DL)) return false; // If the group involves a non-integral pointer, we may not be able to // losslessly cast all values to a common type. unsigned InterleaveFactor = Group->getFactor(); bool ScalarNI = DL.isNonIntegralPointerType(ScalarTy); for (unsigned i = 0; i < InterleaveFactor; i++) { Instruction *Member = Group->getMember(i); if (!Member) continue; auto *MemberTy = getLoadStoreType(Member); bool MemberNI = DL.isNonIntegralPointerType(MemberTy); // Don't coerce non-integral pointers to integers or vice versa. if (MemberNI != ScalarNI) { // TODO: Consider adding special nullptr value case here return false; } else if (MemberNI && ScalarNI && ScalarTy->getPointerAddressSpace() != MemberTy->getPointerAddressSpace()) { return false; } } // Check if masking is required. // A Group may need masking for one of two reasons: it resides in a block that // needs predication, or it was decided to use masking to deal with gaps // (either a gap at the end of a load-access that may result in a speculative // load, or any gaps in a store-access). bool PredicatedAccessRequiresMasking = blockNeedsPredicationForAnyReason(I->getParent()) && Legal->isMaskRequired(I); bool LoadAccessWithGapsRequiresEpilogMasking = isa(I) && Group->requiresScalarEpilogue() && !isScalarEpilogueAllowed(); bool StoreAccessWithGapsRequiresMasking = isa(I) && (Group->getNumMembers() < Group->getFactor()); if (!PredicatedAccessRequiresMasking && !LoadAccessWithGapsRequiresEpilogMasking && !StoreAccessWithGapsRequiresMasking) return true; // If masked interleaving is required, we expect that the user/target had // enabled it, because otherwise it either wouldn't have been created or // it should have been invalidated by the CostModel. assert(useMaskedInterleavedAccesses(TTI) && "Masked interleave-groups for predicated accesses are not enabled."); if (Group->isReverse()) return false; auto *Ty = getLoadStoreType(I); const Align Alignment = getLoadStoreAlignment(I); return isa(I) ? TTI.isLegalMaskedLoad(Ty, Alignment) : TTI.isLegalMaskedStore(Ty, Alignment); } bool LoopVectorizationCostModel::memoryInstructionCanBeWidened( Instruction *I, ElementCount VF) { // Get and ensure we have a valid memory instruction. assert((isa(I)) && "Invalid memory instruction"); auto *Ptr = getLoadStorePointerOperand(I); auto *ScalarTy = getLoadStoreType(I); // In order to be widened, the pointer should be consecutive, first of all. if (!Legal->isConsecutivePtr(ScalarTy, Ptr)) return false; // If the instruction is a store located in a predicated block, it will be // scalarized. if (isScalarWithPredication(I, VF)) return false; // If the instruction's allocated size doesn't equal it's type size, it // requires padding and will be scalarized. auto &DL = I->getModule()->getDataLayout(); if (hasIrregularType(ScalarTy, DL)) return false; return true; } void LoopVectorizationCostModel::collectLoopUniforms(ElementCount VF) { // We should not collect Uniforms more than once per VF. Right now, // this function is called from collectUniformsAndScalars(), which // already does this check. Collecting Uniforms for VF=1 does not make any // sense. assert(VF.isVector() && !Uniforms.contains(VF) && "This function should not be visited twice for the same VF"); // Visit the list of Uniforms. If we'll not find any uniform value, we'll // not analyze again. Uniforms.count(VF) will return 1. Uniforms[VF].clear(); // We now know that the loop is vectorizable! // Collect instructions inside the loop that will remain uniform after // vectorization. // Global values, params and instructions outside of current loop are out of // scope. auto isOutOfScope = [&](Value *V) -> bool { Instruction *I = dyn_cast(V); return (!I || !TheLoop->contains(I)); }; // Worklist containing uniform instructions demanding lane 0. SetVector Worklist; BasicBlock *Latch = TheLoop->getLoopLatch(); // Add uniform instructions demanding lane 0 to the worklist. Instructions // that are scalar with predication must not be considered uniform after // vectorization, because that would create an erroneous replicating region // where only a single instance out of VF should be formed. // TODO: optimize such seldom cases if found important, see PR40816. auto addToWorklistIfAllowed = [&](Instruction *I) -> void { if (isOutOfScope(I)) { LLVM_DEBUG(dbgs() << "LV: Found not uniform due to scope: " << *I << "\n"); return; } if (isScalarWithPredication(I, VF)) { LLVM_DEBUG(dbgs() << "LV: Found not uniform being ScalarWithPredication: " << *I << "\n"); return; } LLVM_DEBUG(dbgs() << "LV: Found uniform instruction: " << *I << "\n"); Worklist.insert(I); }; // Start with the conditional branch. If the branch condition is an // instruction contained in the loop that is only used by the branch, it is // uniform. auto *Cmp = dyn_cast(Latch->getTerminator()->getOperand(0)); if (Cmp && TheLoop->contains(Cmp) && Cmp->hasOneUse()) addToWorklistIfAllowed(Cmp); auto PrevVF = VF.divideCoefficientBy(2); // Return true if all lanes perform the same memory operation, and we can // thus chose to execute only one. auto isUniformMemOpUse = [&](Instruction *I) { // If the value was already known to not be uniform for the previous // (smaller VF), it cannot be uniform for the larger VF. if (PrevVF.isVector()) { auto Iter = Uniforms.find(PrevVF); if (Iter != Uniforms.end() && !Iter->second.contains(I)) return false; } if (!Legal->isUniformMemOp(*I, VF)) return false; if (isa(I)) // Loading the same address always produces the same result - at least // assuming aliasing and ordering which have already been checked. return true; // Storing the same value on every iteration. return TheLoop->isLoopInvariant(cast(I)->getValueOperand()); }; auto isUniformDecision = [&](Instruction *I, ElementCount VF) { InstWidening WideningDecision = getWideningDecision(I, VF); assert(WideningDecision != CM_Unknown && "Widening decision should be ready at this moment"); if (isUniformMemOpUse(I)) return true; return (WideningDecision == CM_Widen || WideningDecision == CM_Widen_Reverse || WideningDecision == CM_Interleave); }; // Returns true if Ptr is the pointer operand of a memory access instruction // I, I is known to not require scalarization, and the pointer is not also // stored. auto isVectorizedMemAccessUse = [&](Instruction *I, Value *Ptr) -> bool { if (isa(I) && I->getOperand(0) == Ptr) return false; return getLoadStorePointerOperand(I) == Ptr && (isUniformDecision(I, VF) || Legal->isInvariant(Ptr)); }; // Holds a list of values which are known to have at least one uniform use. // Note that there may be other uses which aren't uniform. A "uniform use" // here is something which only demands lane 0 of the unrolled iterations; // it does not imply that all lanes produce the same value (e.g. this is not // the usual meaning of uniform) SetVector HasUniformUse; // Scan the loop for instructions which are either a) known to have only // lane 0 demanded or b) are uses which demand only lane 0 of their operand. for (auto *BB : TheLoop->blocks()) for (auto &I : *BB) { if (IntrinsicInst *II = dyn_cast(&I)) { switch (II->getIntrinsicID()) { case Intrinsic::sideeffect: case Intrinsic::experimental_noalias_scope_decl: case Intrinsic::assume: case Intrinsic::lifetime_start: case Intrinsic::lifetime_end: if (TheLoop->hasLoopInvariantOperands(&I)) addToWorklistIfAllowed(&I); break; default: break; } } // ExtractValue instructions must be uniform, because the operands are // known to be loop-invariant. if (auto *EVI = dyn_cast(&I)) { assert(isOutOfScope(EVI->getAggregateOperand()) && "Expected aggregate value to be loop invariant"); addToWorklistIfAllowed(EVI); continue; } // If there's no pointer operand, there's nothing to do. auto *Ptr = getLoadStorePointerOperand(&I); if (!Ptr) continue; if (isUniformMemOpUse(&I)) addToWorklistIfAllowed(&I); if (isVectorizedMemAccessUse(&I, Ptr)) HasUniformUse.insert(Ptr); } // Add to the worklist any operands which have *only* uniform (e.g. lane 0 // demanding) users. Since loops are assumed to be in LCSSA form, this // disallows uses outside the loop as well. for (auto *V : HasUniformUse) { if (isOutOfScope(V)) continue; auto *I = cast(V); auto UsersAreMemAccesses = llvm::all_of(I->users(), [&](User *U) -> bool { return isVectorizedMemAccessUse(cast(U), V); }); if (UsersAreMemAccesses) addToWorklistIfAllowed(I); } // Expand Worklist in topological order: whenever a new instruction // is added , its users should be already inside Worklist. It ensures // a uniform instruction will only be used by uniform instructions. unsigned idx = 0; while (idx != Worklist.size()) { Instruction *I = Worklist[idx++]; for (auto *OV : I->operand_values()) { // isOutOfScope operands cannot be uniform instructions. if (isOutOfScope(OV)) continue; // First order recurrence Phi's should typically be considered // non-uniform. auto *OP = dyn_cast(OV); if (OP && Legal->isFixedOrderRecurrence(OP)) continue; // If all the users of the operand are uniform, then add the // operand into the uniform worklist. auto *OI = cast(OV); if (llvm::all_of(OI->users(), [&](User *U) -> bool { auto *J = cast(U); return Worklist.count(J) || isVectorizedMemAccessUse(J, OI); })) addToWorklistIfAllowed(OI); } } // For an instruction to be added into Worklist above, all its users inside // the loop should also be in Worklist. However, this condition cannot be // true for phi nodes that form a cyclic dependence. We must process phi // nodes separately. An induction variable will remain uniform if all users // of the induction variable and induction variable update remain uniform. // The code below handles both pointer and non-pointer induction variables. for (const auto &Induction : Legal->getInductionVars()) { auto *Ind = Induction.first; auto *IndUpdate = cast(Ind->getIncomingValueForBlock(Latch)); // Determine if all users of the induction variable are uniform after // vectorization. auto UniformInd = llvm::all_of(Ind->users(), [&](User *U) -> bool { auto *I = cast(U); return I == IndUpdate || !TheLoop->contains(I) || Worklist.count(I) || isVectorizedMemAccessUse(I, Ind); }); if (!UniformInd) continue; // Determine if all users of the induction variable update instruction are // uniform after vectorization. auto UniformIndUpdate = llvm::all_of(IndUpdate->users(), [&](User *U) -> bool { auto *I = cast(U); return I == Ind || !TheLoop->contains(I) || Worklist.count(I) || isVectorizedMemAccessUse(I, IndUpdate); }); if (!UniformIndUpdate) continue; // The induction variable and its update instruction will remain uniform. addToWorklistIfAllowed(Ind); addToWorklistIfAllowed(IndUpdate); } Uniforms[VF].insert(Worklist.begin(), Worklist.end()); } bool LoopVectorizationCostModel::runtimeChecksRequired() { LLVM_DEBUG(dbgs() << "LV: Performing code size checks.\n"); if (Legal->getRuntimePointerChecking()->Need) { reportVectorizationFailure("Runtime ptr check is required with -Os/-Oz", "runtime pointer checks needed. Enable vectorization of this " "loop with '#pragma clang loop vectorize(enable)' when " "compiling with -Os/-Oz", "CantVersionLoopWithOptForSize", ORE, TheLoop); return true; } if (!PSE.getPredicate().isAlwaysTrue()) { reportVectorizationFailure("Runtime SCEV check is required with -Os/-Oz", "runtime SCEV checks needed. Enable vectorization of this " "loop with '#pragma clang loop vectorize(enable)' when " "compiling with -Os/-Oz", "CantVersionLoopWithOptForSize", ORE, TheLoop); return true; } // FIXME: Avoid specializing for stride==1 instead of bailing out. if (!Legal->getLAI()->getSymbolicStrides().empty()) { reportVectorizationFailure("Runtime stride check for small trip count", "runtime stride == 1 checks needed. Enable vectorization of " "this loop without such check by compiling with -Os/-Oz", "CantVersionLoopWithOptForSize", ORE, TheLoop); return true; } return false; } ElementCount LoopVectorizationCostModel::getMaxLegalScalableVF(unsigned MaxSafeElements) { if (!TTI.supportsScalableVectors() && !ForceTargetSupportsScalableVectors) return ElementCount::getScalable(0); if (Hints->isScalableVectorizationDisabled()) { reportVectorizationInfo("Scalable vectorization is explicitly disabled", "ScalableVectorizationDisabled", ORE, TheLoop); return ElementCount::getScalable(0); } LLVM_DEBUG(dbgs() << "LV: Scalable vectorization is available\n"); auto MaxScalableVF = ElementCount::getScalable( std::numeric_limits::max()); // Test that the loop-vectorizer can legalize all operations for this MaxVF. // FIXME: While for scalable vectors this is currently sufficient, this should // be replaced by a more detailed mechanism that filters out specific VFs, // instead of invalidating vectorization for a whole set of VFs based on the // MaxVF. // Disable scalable vectorization if the loop contains unsupported reductions. if (!canVectorizeReductions(MaxScalableVF)) { reportVectorizationInfo( "Scalable vectorization not supported for the reduction " "operations found in this loop.", "ScalableVFUnfeasible", ORE, TheLoop); return ElementCount::getScalable(0); } // Disable scalable vectorization if the loop contains any instructions // with element types not supported for scalable vectors. if (any_of(ElementTypesInLoop, [&](Type *Ty) { return !Ty->isVoidTy() && !this->TTI.isElementTypeLegalForScalableVector(Ty); })) { reportVectorizationInfo("Scalable vectorization is not supported " "for all element types found in this loop.", "ScalableVFUnfeasible", ORE, TheLoop); return ElementCount::getScalable(0); } if (Legal->isSafeForAnyVectorWidth()) return MaxScalableVF; // Limit MaxScalableVF by the maximum safe dependence distance. if (std::optional MaxVScale = getMaxVScale(*TheFunction, TTI)) MaxScalableVF = ElementCount::getScalable(MaxSafeElements / *MaxVScale); else MaxScalableVF = ElementCount::getScalable(0); if (!MaxScalableVF) reportVectorizationInfo( "Max legal vector width too small, scalable vectorization " "unfeasible.", "ScalableVFUnfeasible", ORE, TheLoop); return MaxScalableVF; } FixedScalableVFPair LoopVectorizationCostModel::computeFeasibleMaxVF( unsigned MaxTripCount, ElementCount UserVF, bool FoldTailByMasking) { MinBWs = computeMinimumValueSizes(TheLoop->getBlocks(), *DB, &TTI); unsigned SmallestType, WidestType; std::tie(SmallestType, WidestType) = getSmallestAndWidestTypes(); // Get the maximum safe dependence distance in bits computed by LAA. // It is computed by MaxVF * sizeOf(type) * 8, where type is taken from // the memory accesses that is most restrictive (involved in the smallest // dependence distance). unsigned MaxSafeElements = llvm::bit_floor(Legal->getMaxSafeVectorWidthInBits() / WidestType); auto MaxSafeFixedVF = ElementCount::getFixed(MaxSafeElements); auto MaxSafeScalableVF = getMaxLegalScalableVF(MaxSafeElements); LLVM_DEBUG(dbgs() << "LV: The max safe fixed VF is: " << MaxSafeFixedVF << ".\n"); LLVM_DEBUG(dbgs() << "LV: The max safe scalable VF is: " << MaxSafeScalableVF << ".\n"); // First analyze the UserVF, fall back if the UserVF should be ignored. if (UserVF) { auto MaxSafeUserVF = UserVF.isScalable() ? MaxSafeScalableVF : MaxSafeFixedVF; if (ElementCount::isKnownLE(UserVF, MaxSafeUserVF)) { // If `VF=vscale x N` is safe, then so is `VF=N` if (UserVF.isScalable()) return FixedScalableVFPair( ElementCount::getFixed(UserVF.getKnownMinValue()), UserVF); else return UserVF; } assert(ElementCount::isKnownGT(UserVF, MaxSafeUserVF)); // Only clamp if the UserVF is not scalable. If the UserVF is scalable, it // is better to ignore the hint and let the compiler choose a suitable VF. if (!UserVF.isScalable()) { LLVM_DEBUG(dbgs() << "LV: User VF=" << UserVF << " is unsafe, clamping to max safe VF=" << MaxSafeFixedVF << ".\n"); ORE->emit([&]() { return OptimizationRemarkAnalysis(DEBUG_TYPE, "VectorizationFactor", TheLoop->getStartLoc(), TheLoop->getHeader()) << "User-specified vectorization factor " << ore::NV("UserVectorizationFactor", UserVF) << " is unsafe, clamping to maximum safe vectorization factor " << ore::NV("VectorizationFactor", MaxSafeFixedVF); }); return MaxSafeFixedVF; } if (!TTI.supportsScalableVectors() && !ForceTargetSupportsScalableVectors) { LLVM_DEBUG(dbgs() << "LV: User VF=" << UserVF << " is ignored because scalable vectors are not " "available.\n"); ORE->emit([&]() { return OptimizationRemarkAnalysis(DEBUG_TYPE, "VectorizationFactor", TheLoop->getStartLoc(), TheLoop->getHeader()) << "User-specified vectorization factor " << ore::NV("UserVectorizationFactor", UserVF) << " is ignored because the target does not support scalable " "vectors. The compiler will pick a more suitable value."; }); } else { LLVM_DEBUG(dbgs() << "LV: User VF=" << UserVF << " is unsafe. Ignoring scalable UserVF.\n"); ORE->emit([&]() { return OptimizationRemarkAnalysis(DEBUG_TYPE, "VectorizationFactor", TheLoop->getStartLoc(), TheLoop->getHeader()) << "User-specified vectorization factor " << ore::NV("UserVectorizationFactor", UserVF) << " is unsafe. Ignoring the hint to let the compiler pick a " "more suitable value."; }); } } LLVM_DEBUG(dbgs() << "LV: The Smallest and Widest types: " << SmallestType << " / " << WidestType << " bits.\n"); FixedScalableVFPair Result(ElementCount::getFixed(1), ElementCount::getScalable(0)); if (auto MaxVF = getMaximizedVFForTarget(MaxTripCount, SmallestType, WidestType, MaxSafeFixedVF, FoldTailByMasking)) Result.FixedVF = MaxVF; if (auto MaxVF = getMaximizedVFForTarget(MaxTripCount, SmallestType, WidestType, MaxSafeScalableVF, FoldTailByMasking)) if (MaxVF.isScalable()) { Result.ScalableVF = MaxVF; LLVM_DEBUG(dbgs() << "LV: Found feasible scalable VF = " << MaxVF << "\n"); } return Result; } FixedScalableVFPair LoopVectorizationCostModel::computeMaxVF(ElementCount UserVF, unsigned UserIC) { if (Legal->getRuntimePointerChecking()->Need && TTI.hasBranchDivergence()) { // TODO: It may by useful to do since it's still likely to be dynamically // uniform if the target can skip. reportVectorizationFailure( "Not inserting runtime ptr check for divergent target", "runtime pointer checks needed. Not enabled for divergent target", "CantVersionLoopWithDivergentTarget", ORE, TheLoop); return FixedScalableVFPair::getNone(); } unsigned TC = PSE.getSE()->getSmallConstantTripCount(TheLoop); unsigned MaxTC = PSE.getSE()->getSmallConstantMaxTripCount(TheLoop); LLVM_DEBUG(dbgs() << "LV: Found trip count: " << TC << '\n'); if (TC == 1) { reportVectorizationFailure("Single iteration (non) loop", "loop trip count is one, irrelevant for vectorization", "SingleIterationLoop", ORE, TheLoop); return FixedScalableVFPair::getNone(); } switch (ScalarEpilogueStatus) { case CM_ScalarEpilogueAllowed: return computeFeasibleMaxVF(MaxTC, UserVF, false); case CM_ScalarEpilogueNotAllowedUsePredicate: [[fallthrough]]; case CM_ScalarEpilogueNotNeededUsePredicate: LLVM_DEBUG( dbgs() << "LV: vector predicate hint/switch found.\n" << "LV: Not allowing scalar epilogue, creating predicated " << "vector loop.\n"); break; case CM_ScalarEpilogueNotAllowedLowTripLoop: // fallthrough as a special case of OptForSize case CM_ScalarEpilogueNotAllowedOptSize: if (ScalarEpilogueStatus == CM_ScalarEpilogueNotAllowedOptSize) LLVM_DEBUG( dbgs() << "LV: Not allowing scalar epilogue due to -Os/-Oz.\n"); else LLVM_DEBUG(dbgs() << "LV: Not allowing scalar epilogue due to low trip " << "count.\n"); // Bail if runtime checks are required, which are not good when optimising // for size. if (runtimeChecksRequired()) return FixedScalableVFPair::getNone(); break; } // The only loops we can vectorize without a scalar epilogue, are loops with // a bottom-test and a single exiting block. We'd have to handle the fact // that not every instruction executes on the last iteration. This will // require a lane mask which varies through the vector loop body. (TODO) if (TheLoop->getExitingBlock() != TheLoop->getLoopLatch()) { // If there was a tail-folding hint/switch, but we can't fold the tail by // masking, fallback to a vectorization with a scalar epilogue. if (ScalarEpilogueStatus == CM_ScalarEpilogueNotNeededUsePredicate) { LLVM_DEBUG(dbgs() << "LV: Cannot fold tail by masking: vectorize with a " "scalar epilogue instead.\n"); ScalarEpilogueStatus = CM_ScalarEpilogueAllowed; return computeFeasibleMaxVF(MaxTC, UserVF, false); } return FixedScalableVFPair::getNone(); } // Now try the tail folding // Invalidate interleave groups that require an epilogue if we can't mask // the interleave-group. if (!useMaskedInterleavedAccesses(TTI)) { assert(WideningDecisions.empty() && Uniforms.empty() && Scalars.empty() && "No decisions should have been taken at this point"); // Note: There is no need to invalidate any cost modeling decisions here, as // non where taken so far. InterleaveInfo.invalidateGroupsRequiringScalarEpilogue(); } FixedScalableVFPair MaxFactors = computeFeasibleMaxVF(MaxTC, UserVF, true); // Avoid tail folding if the trip count is known to be a multiple of any VF // we choose. std::optional MaxPowerOf2RuntimeVF = MaxFactors.FixedVF.getFixedValue(); if (MaxFactors.ScalableVF) { std::optional MaxVScale = getMaxVScale(*TheFunction, TTI); if (MaxVScale && TTI.isVScaleKnownToBeAPowerOfTwo()) { MaxPowerOf2RuntimeVF = std::max( *MaxPowerOf2RuntimeVF, *MaxVScale * MaxFactors.ScalableVF.getKnownMinValue()); } else MaxPowerOf2RuntimeVF = std::nullopt; // Stick with tail-folding for now. } if (MaxPowerOf2RuntimeVF && *MaxPowerOf2RuntimeVF > 0) { assert((UserVF.isNonZero() || isPowerOf2_32(*MaxPowerOf2RuntimeVF)) && "MaxFixedVF must be a power of 2"); unsigned MaxVFtimesIC = UserIC ? *MaxPowerOf2RuntimeVF * UserIC : *MaxPowerOf2RuntimeVF; ScalarEvolution *SE = PSE.getSE(); const SCEV *BackedgeTakenCount = PSE.getBackedgeTakenCount(); const SCEV *ExitCount = SE->getAddExpr( BackedgeTakenCount, SE->getOne(BackedgeTakenCount->getType())); const SCEV *Rem = SE->getURemExpr( SE->applyLoopGuards(ExitCount, TheLoop), SE->getConstant(BackedgeTakenCount->getType(), MaxVFtimesIC)); if (Rem->isZero()) { // Accept MaxFixedVF if we do not have a tail. LLVM_DEBUG(dbgs() << "LV: No tail will remain for any chosen VF.\n"); return MaxFactors; } } // If we don't know the precise trip count, or if the trip count that we // found modulo the vectorization factor is not zero, try to fold the tail // by masking. // FIXME: look for a smaller MaxVF that does divide TC rather than masking. if (Legal->prepareToFoldTailByMasking()) { CanFoldTailByMasking = true; return MaxFactors; } // If there was a tail-folding hint/switch, but we can't fold the tail by // masking, fallback to a vectorization with a scalar epilogue. if (ScalarEpilogueStatus == CM_ScalarEpilogueNotNeededUsePredicate) { LLVM_DEBUG(dbgs() << "LV: Cannot fold tail by masking: vectorize with a " "scalar epilogue instead.\n"); ScalarEpilogueStatus = CM_ScalarEpilogueAllowed; return MaxFactors; } if (ScalarEpilogueStatus == CM_ScalarEpilogueNotAllowedUsePredicate) { LLVM_DEBUG(dbgs() << "LV: Can't fold tail by masking: don't vectorize\n"); return FixedScalableVFPair::getNone(); } if (TC == 0) { reportVectorizationFailure( "Unable to calculate the loop count due to complex control flow", "unable to calculate the loop count due to complex control flow", "UnknownLoopCountComplexCFG", ORE, TheLoop); return FixedScalableVFPair::getNone(); } reportVectorizationFailure( "Cannot optimize for size and vectorize at the same time.", "cannot optimize for size and vectorize at the same time. " "Enable vectorization of this loop with '#pragma clang loop " "vectorize(enable)' when compiling with -Os/-Oz", "NoTailLoopWithOptForSize", ORE, TheLoop); return FixedScalableVFPair::getNone(); } ElementCount LoopVectorizationCostModel::getMaximizedVFForTarget( unsigned MaxTripCount, unsigned SmallestType, unsigned WidestType, ElementCount MaxSafeVF, bool FoldTailByMasking) { bool ComputeScalableMaxVF = MaxSafeVF.isScalable(); const TypeSize WidestRegister = TTI.getRegisterBitWidth( ComputeScalableMaxVF ? TargetTransformInfo::RGK_ScalableVector : TargetTransformInfo::RGK_FixedWidthVector); // Convenience function to return the minimum of two ElementCounts. auto MinVF = [](const ElementCount &LHS, const ElementCount &RHS) { assert((LHS.isScalable() == RHS.isScalable()) && "Scalable flags must match"); return ElementCount::isKnownLT(LHS, RHS) ? LHS : RHS; }; // Ensure MaxVF is a power of 2; the dependence distance bound may not be. // Note that both WidestRegister and WidestType may not be a powers of 2. auto MaxVectorElementCount = ElementCount::get( llvm::bit_floor(WidestRegister.getKnownMinValue() / WidestType), ComputeScalableMaxVF); MaxVectorElementCount = MinVF(MaxVectorElementCount, MaxSafeVF); LLVM_DEBUG(dbgs() << "LV: The Widest register safe to use is: " << (MaxVectorElementCount * WidestType) << " bits.\n"); if (!MaxVectorElementCount) { LLVM_DEBUG(dbgs() << "LV: The target has no " << (ComputeScalableMaxVF ? "scalable" : "fixed") << " vector registers.\n"); return ElementCount::getFixed(1); } unsigned WidestRegisterMinEC = MaxVectorElementCount.getKnownMinValue(); if (MaxVectorElementCount.isScalable() && TheFunction->hasFnAttribute(Attribute::VScaleRange)) { auto Attr = TheFunction->getFnAttribute(Attribute::VScaleRange); auto Min = Attr.getVScaleRangeMin(); WidestRegisterMinEC *= Min; } // When a scalar epilogue is required, at least one iteration of the scalar // loop has to execute. Adjust MaxTripCount accordingly to avoid picking a // max VF that results in a dead vector loop. if (MaxTripCount > 0 && requiresScalarEpilogue(true)) MaxTripCount -= 1; if (MaxTripCount && MaxTripCount <= WidestRegisterMinEC && (!FoldTailByMasking || isPowerOf2_32(MaxTripCount))) { // If upper bound loop trip count (TC) is known at compile time there is no // point in choosing VF greater than TC (as done in the loop below). Select // maximum power of two which doesn't exceed TC. If MaxVectorElementCount is // scalable, we only fall back on a fixed VF when the TC is less than or // equal to the known number of lanes. auto ClampedUpperTripCount = llvm::bit_floor(MaxTripCount); LLVM_DEBUG(dbgs() << "LV: Clamping the MaxVF to maximum power of two not " "exceeding the constant trip count: " << ClampedUpperTripCount << "\n"); return ElementCount::get( ClampedUpperTripCount, FoldTailByMasking ? MaxVectorElementCount.isScalable() : false); } TargetTransformInfo::RegisterKind RegKind = ComputeScalableMaxVF ? TargetTransformInfo::RGK_ScalableVector : TargetTransformInfo::RGK_FixedWidthVector; ElementCount MaxVF = MaxVectorElementCount; if (MaximizeBandwidth || (MaximizeBandwidth.getNumOccurrences() == 0 && (TTI.shouldMaximizeVectorBandwidth(RegKind) || (UseWiderVFIfCallVariantsPresent && Legal->hasVectorCallVariants())))) { auto MaxVectorElementCountMaxBW = ElementCount::get( llvm::bit_floor(WidestRegister.getKnownMinValue() / SmallestType), ComputeScalableMaxVF); MaxVectorElementCountMaxBW = MinVF(MaxVectorElementCountMaxBW, MaxSafeVF); // Collect all viable vectorization factors larger than the default MaxVF // (i.e. MaxVectorElementCount). SmallVector VFs; for (ElementCount VS = MaxVectorElementCount * 2; ElementCount::isKnownLE(VS, MaxVectorElementCountMaxBW); VS *= 2) VFs.push_back(VS); // For each VF calculate its register usage. auto RUs = calculateRegisterUsage(VFs); // Select the largest VF which doesn't require more registers than existing // ones. for (int i = RUs.size() - 1; i >= 0; --i) { bool Selected = true; for (auto &pair : RUs[i].MaxLocalUsers) { unsigned TargetNumRegisters = TTI.getNumberOfRegisters(pair.first); if (pair.second > TargetNumRegisters) Selected = false; } if (Selected) { MaxVF = VFs[i]; break; } } if (ElementCount MinVF = TTI.getMinimumVF(SmallestType, ComputeScalableMaxVF)) { if (ElementCount::isKnownLT(MaxVF, MinVF)) { LLVM_DEBUG(dbgs() << "LV: Overriding calculated MaxVF(" << MaxVF << ") with target's minimum: " << MinVF << '\n'); MaxVF = MinVF; } } // Invalidate any widening decisions we might have made, in case the loop // requires prediction (decided later), but we have already made some // load/store widening decisions. invalidateCostModelingDecisions(); } return MaxVF; } /// Convenience function that returns the value of vscale_range iff /// vscale_range.min == vscale_range.max or otherwise returns the value /// returned by the corresponding TTI method. static std::optional getVScaleForTuning(const Loop *L, const TargetTransformInfo &TTI) { const Function *Fn = L->getHeader()->getParent(); if (Fn->hasFnAttribute(Attribute::VScaleRange)) { auto Attr = Fn->getFnAttribute(Attribute::VScaleRange); auto Min = Attr.getVScaleRangeMin(); auto Max = Attr.getVScaleRangeMax(); if (Max && Min == Max) return Max; } return TTI.getVScaleForTuning(); } bool LoopVectorizationPlanner::isMoreProfitable( const VectorizationFactor &A, const VectorizationFactor &B) const { InstructionCost CostA = A.Cost; InstructionCost CostB = B.Cost; unsigned MaxTripCount = PSE.getSE()->getSmallConstantMaxTripCount(OrigLoop); if (!A.Width.isScalable() && !B.Width.isScalable() && MaxTripCount) { // If the trip count is a known (possibly small) constant, the trip count // will be rounded up to an integer number of iterations under // FoldTailByMasking. The total cost in that case will be // VecCost*ceil(TripCount/VF). When not folding the tail, the total // cost will be VecCost*floor(TC/VF) + ScalarCost*(TC%VF). There will be // some extra overheads, but for the purpose of comparing the costs of // different VFs we can use this to compare the total loop-body cost // expected after vectorization. auto GetCostForTC = [MaxTripCount, this](unsigned VF, InstructionCost VectorCost, InstructionCost ScalarCost) { return CM.foldTailByMasking() ? VectorCost * divideCeil(MaxTripCount, VF) : VectorCost * (MaxTripCount / VF) + ScalarCost * (MaxTripCount % VF); }; auto RTCostA = GetCostForTC(A.Width.getFixedValue(), CostA, A.ScalarCost); auto RTCostB = GetCostForTC(B.Width.getFixedValue(), CostB, B.ScalarCost); return RTCostA < RTCostB; } // Improve estimate for the vector width if it is scalable. unsigned EstimatedWidthA = A.Width.getKnownMinValue(); unsigned EstimatedWidthB = B.Width.getKnownMinValue(); if (std::optional VScale = getVScaleForTuning(OrigLoop, TTI)) { if (A.Width.isScalable()) EstimatedWidthA *= *VScale; if (B.Width.isScalable()) EstimatedWidthB *= *VScale; } // Assume vscale may be larger than 1 (or the value being tuned for), // so that scalable vectorization is slightly favorable over fixed-width // vectorization. if (A.Width.isScalable() && !B.Width.isScalable()) return (CostA * B.Width.getFixedValue()) <= (CostB * EstimatedWidthA); // To avoid the need for FP division: // (CostA / A.Width) < (CostB / B.Width) // <=> (CostA * B.Width) < (CostB * A.Width) return (CostA * EstimatedWidthB) < (CostB * EstimatedWidthA); } static void emitInvalidCostRemarks(SmallVector InvalidCosts, OptimizationRemarkEmitter *ORE, Loop *TheLoop) { if (InvalidCosts.empty()) return; // Emit a report of VFs with invalid costs in the loop. // Group the remarks per instruction, keeping the instruction order from // InvalidCosts. std::map Numbering; unsigned I = 0; for (auto &Pair : InvalidCosts) if (!Numbering.count(Pair.first)) Numbering[Pair.first] = I++; // Sort the list, first on instruction(number) then on VF. sort(InvalidCosts, [&Numbering](InstructionVFPair &A, InstructionVFPair &B) { if (Numbering[A.first] != Numbering[B.first]) return Numbering[A.first] < Numbering[B.first]; ElementCountComparator ECC; return ECC(A.second, B.second); }); // For a list of ordered instruction-vf pairs: // [(load, vf1), (load, vf2), (store, vf1)] // Group the instructions together to emit separate remarks for: // load (vf1, vf2) // store (vf1) auto Tail = ArrayRef(InvalidCosts); auto Subset = ArrayRef(); do { if (Subset.empty()) Subset = Tail.take_front(1); Instruction *I = Subset.front().first; // If the next instruction is different, or if there are no other pairs, // emit a remark for the collated subset. e.g. // [(load, vf1), (load, vf2))] // to emit: // remark: invalid costs for 'load' at VF=(vf, vf2) if (Subset == Tail || Tail[Subset.size()].first != I) { std::string OutString; raw_string_ostream OS(OutString); assert(!Subset.empty() && "Unexpected empty range"); OS << "Instruction with invalid costs prevented vectorization at VF=("; for (const auto &Pair : Subset) OS << (Pair.second == Subset.front().second ? "" : ", ") << Pair.second; OS << "):"; if (auto *CI = dyn_cast(I)) OS << " call to " << CI->getCalledFunction()->getName(); else OS << " " << I->getOpcodeName(); OS.flush(); reportVectorizationInfo(OutString, "InvalidCost", ORE, TheLoop, I); Tail = Tail.drop_front(Subset.size()); Subset = {}; } else // Grow the subset by one element Subset = Tail.take_front(Subset.size() + 1); } while (!Tail.empty()); } VectorizationFactor LoopVectorizationPlanner::selectVectorizationFactor( const ElementCountSet &VFCandidates) { InstructionCost ExpectedCost = CM.expectedCost(ElementCount::getFixed(1)).first; LLVM_DEBUG(dbgs() << "LV: Scalar loop costs: " << ExpectedCost << ".\n"); assert(ExpectedCost.isValid() && "Unexpected invalid cost for scalar loop"); assert(VFCandidates.count(ElementCount::getFixed(1)) && "Expected Scalar VF to be a candidate"); const VectorizationFactor ScalarCost(ElementCount::getFixed(1), ExpectedCost, ExpectedCost); VectorizationFactor ChosenFactor = ScalarCost; bool ForceVectorization = Hints.getForce() == LoopVectorizeHints::FK_Enabled; if (ForceVectorization && VFCandidates.size() > 1) { // Ignore scalar width, because the user explicitly wants vectorization. // Initialize cost to max so that VF = 2 is, at least, chosen during cost // evaluation. ChosenFactor.Cost = InstructionCost::getMax(); } SmallVector InvalidCosts; for (const auto &i : VFCandidates) { // The cost for scalar VF=1 is already calculated, so ignore it. if (i.isScalar()) continue; LoopVectorizationCostModel::VectorizationCostTy C = CM.expectedCost(i, &InvalidCosts); VectorizationFactor Candidate(i, C.first, ScalarCost.ScalarCost); #ifndef NDEBUG unsigned AssumedMinimumVscale = getVScaleForTuning(OrigLoop, TTI).value_or(1); unsigned Width = Candidate.Width.isScalable() ? Candidate.Width.getKnownMinValue() * AssumedMinimumVscale : Candidate.Width.getFixedValue(); LLVM_DEBUG(dbgs() << "LV: Vector loop of width " << i << " costs: " << (Candidate.Cost / Width)); if (i.isScalable()) LLVM_DEBUG(dbgs() << " (assuming a minimum vscale of " << AssumedMinimumVscale << ")"); LLVM_DEBUG(dbgs() << ".\n"); #endif if (!C.second && !ForceVectorization) { LLVM_DEBUG( dbgs() << "LV: Not considering vector loop of width " << i << " because it will not generate any vector instructions.\n"); continue; } // If profitable add it to ProfitableVF list. if (isMoreProfitable(Candidate, ScalarCost)) ProfitableVFs.push_back(Candidate); if (isMoreProfitable(Candidate, ChosenFactor)) ChosenFactor = Candidate; } emitInvalidCostRemarks(InvalidCosts, ORE, OrigLoop); if (!EnableCondStoresVectorization && CM.hasPredStores()) { reportVectorizationFailure( "There are conditional stores.", "store that is conditionally executed prevents vectorization", "ConditionalStore", ORE, OrigLoop); ChosenFactor = ScalarCost; } LLVM_DEBUG(if (ForceVectorization && !ChosenFactor.Width.isScalar() && !isMoreProfitable(ChosenFactor, ScalarCost)) dbgs() << "LV: Vectorization seems to be not beneficial, " << "but was forced by a user.\n"); LLVM_DEBUG(dbgs() << "LV: Selecting VF: " << ChosenFactor.Width << ".\n"); return ChosenFactor; } bool LoopVectorizationPlanner::isCandidateForEpilogueVectorization( ElementCount VF) const { // Cross iteration phis such as reductions need special handling and are // currently unsupported. if (any_of(OrigLoop->getHeader()->phis(), [&](PHINode &Phi) { return Legal->isFixedOrderRecurrence(&Phi); })) return false; // Phis with uses outside of the loop require special handling and are // currently unsupported. for (const auto &Entry : Legal->getInductionVars()) { // Look for uses of the value of the induction at the last iteration. Value *PostInc = Entry.first->getIncomingValueForBlock(OrigLoop->getLoopLatch()); for (User *U : PostInc->users()) if (!OrigLoop->contains(cast(U))) return false; // Look for uses of penultimate value of the induction. for (User *U : Entry.first->users()) if (!OrigLoop->contains(cast(U))) return false; } // Epilogue vectorization code has not been auditted to ensure it handles // non-latch exits properly. It may be fine, but it needs auditted and // tested. if (OrigLoop->getExitingBlock() != OrigLoop->getLoopLatch()) return false; return true; } bool LoopVectorizationCostModel::isEpilogueVectorizationProfitable( const ElementCount VF) const { // FIXME: We need a much better cost-model to take different parameters such // as register pressure, code size increase and cost of extra branches into // account. For now we apply a very crude heuristic and only consider loops // with vectorization factors larger than a certain value. // Allow the target to opt out entirely. if (!TTI.preferEpilogueVectorization()) return false; // We also consider epilogue vectorization unprofitable for targets that don't // consider interleaving beneficial (eg. MVE). if (TTI.getMaxInterleaveFactor(VF) <= 1) return false; unsigned Multiplier = 1; if (VF.isScalable()) Multiplier = getVScaleForTuning(TheLoop, TTI).value_or(1); if ((Multiplier * VF.getKnownMinValue()) >= EpilogueVectorizationMinVF) return true; return false; } VectorizationFactor LoopVectorizationPlanner::selectEpilogueVectorizationFactor( const ElementCount MainLoopVF, unsigned IC) { VectorizationFactor Result = VectorizationFactor::Disabled(); if (!EnableEpilogueVectorization) { LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization is disabled.\n"); return Result; } if (!CM.isScalarEpilogueAllowed()) { LLVM_DEBUG(dbgs() << "LEV: Unable to vectorize epilogue because no " "epilogue is allowed.\n"); return Result; } // Not really a cost consideration, but check for unsupported cases here to // simplify the logic. if (!isCandidateForEpilogueVectorization(MainLoopVF)) { LLVM_DEBUG(dbgs() << "LEV: Unable to vectorize epilogue because the loop " "is not a supported candidate.\n"); return Result; } if (EpilogueVectorizationForceVF > 1) { LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization factor is forced.\n"); ElementCount ForcedEC = ElementCount::getFixed(EpilogueVectorizationForceVF); if (hasPlanWithVF(ForcedEC)) return {ForcedEC, 0, 0}; else { LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization forced factor is not " "viable.\n"); return Result; } } if (OrigLoop->getHeader()->getParent()->hasOptSize() || OrigLoop->getHeader()->getParent()->hasMinSize()) { LLVM_DEBUG( dbgs() << "LEV: Epilogue vectorization skipped due to opt for size.\n"); return Result; } if (!CM.isEpilogueVectorizationProfitable(MainLoopVF)) { LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization is not profitable for " "this loop\n"); return Result; } // If MainLoopVF = vscale x 2, and vscale is expected to be 4, then we know // the main loop handles 8 lanes per iteration. We could still benefit from // vectorizing the epilogue loop with VF=4. ElementCount EstimatedRuntimeVF = MainLoopVF; if (MainLoopVF.isScalable()) { EstimatedRuntimeVF = ElementCount::getFixed(MainLoopVF.getKnownMinValue()); if (std::optional VScale = getVScaleForTuning(OrigLoop, TTI)) EstimatedRuntimeVF *= *VScale; } ScalarEvolution &SE = *PSE.getSE(); Type *TCType = Legal->getWidestInductionType(); const SCEV *RemainingIterations = nullptr; for (auto &NextVF : ProfitableVFs) { // Skip candidate VFs without a corresponding VPlan. if (!hasPlanWithVF(NextVF.Width)) continue; // Skip candidate VFs with widths >= the estimate runtime VF (scalable // vectors) or the VF of the main loop (fixed vectors). if ((!NextVF.Width.isScalable() && MainLoopVF.isScalable() && ElementCount::isKnownGE(NextVF.Width, EstimatedRuntimeVF)) || ElementCount::isKnownGE(NextVF.Width, MainLoopVF)) continue; // If NextVF is greater than the number of remaining iterations, the // epilogue loop would be dead. Skip such factors. if (!MainLoopVF.isScalable() && !NextVF.Width.isScalable()) { // TODO: extend to support scalable VFs. if (!RemainingIterations) { const SCEV *TC = createTripCountSCEV(TCType, PSE, OrigLoop); RemainingIterations = SE.getURemExpr( TC, SE.getConstant(TCType, MainLoopVF.getKnownMinValue() * IC)); } if (SE.isKnownPredicate( CmpInst::ICMP_UGT, SE.getConstant(TCType, NextVF.Width.getKnownMinValue()), RemainingIterations)) continue; } if (Result.Width.isScalar() || isMoreProfitable(NextVF, Result)) Result = NextVF; } if (Result != VectorizationFactor::Disabled()) LLVM_DEBUG(dbgs() << "LEV: Vectorizing epilogue loop with VF = " << Result.Width << "\n"); return Result; } std::pair LoopVectorizationCostModel::getSmallestAndWidestTypes() { unsigned MinWidth = -1U; unsigned MaxWidth = 8; const DataLayout &DL = TheFunction->getParent()->getDataLayout(); // For in-loop reductions, no element types are added to ElementTypesInLoop // if there are no loads/stores in the loop. In this case, check through the // reduction variables to determine the maximum width. if (ElementTypesInLoop.empty() && !Legal->getReductionVars().empty()) { // Reset MaxWidth so that we can find the smallest type used by recurrences // in the loop. MaxWidth = -1U; for (const auto &PhiDescriptorPair : Legal->getReductionVars()) { const RecurrenceDescriptor &RdxDesc = PhiDescriptorPair.second; // When finding the min width used by the recurrence we need to account // for casts on the input operands of the recurrence. MaxWidth = std::min( MaxWidth, std::min( RdxDesc.getMinWidthCastToRecurrenceTypeInBits(), RdxDesc.getRecurrenceType()->getScalarSizeInBits())); } } else { for (Type *T : ElementTypesInLoop) { MinWidth = std::min( MinWidth, DL.getTypeSizeInBits(T->getScalarType()).getFixedValue()); MaxWidth = std::max( MaxWidth, DL.getTypeSizeInBits(T->getScalarType()).getFixedValue()); } } return {MinWidth, MaxWidth}; } void LoopVectorizationCostModel::collectElementTypesForWidening() { ElementTypesInLoop.clear(); // For each block. for (BasicBlock *BB : TheLoop->blocks()) { // For each instruction in the loop. for (Instruction &I : BB->instructionsWithoutDebug()) { Type *T = I.getType(); // Skip ignored values. if (ValuesToIgnore.count(&I)) continue; // Only examine Loads, Stores and PHINodes. if (!isa(I) && !isa(I) && !isa(I)) continue; // Examine PHI nodes that are reduction variables. Update the type to // account for the recurrence type. if (auto *PN = dyn_cast(&I)) { if (!Legal->isReductionVariable(PN)) continue; const RecurrenceDescriptor &RdxDesc = Legal->getReductionVars().find(PN)->second; if (PreferInLoopReductions || useOrderedReductions(RdxDesc) || TTI.preferInLoopReduction(RdxDesc.getOpcode(), RdxDesc.getRecurrenceType(), TargetTransformInfo::ReductionFlags())) continue; T = RdxDesc.getRecurrenceType(); } // Examine the stored values. if (auto *ST = dyn_cast(&I)) T = ST->getValueOperand()->getType(); assert(T->isSized() && "Expected the load/store/recurrence type to be sized"); ElementTypesInLoop.insert(T); } } } unsigned LoopVectorizationCostModel::selectInterleaveCount(ElementCount VF, InstructionCost LoopCost) { // -- The interleave heuristics -- // We interleave the loop in order to expose ILP and reduce the loop overhead. // There are many micro-architectural considerations that we can't predict // at this level. For example, frontend pressure (on decode or fetch) due to // code size, or the number and capabilities of the execution ports. // // We use the following heuristics to select the interleave count: // 1. If the code has reductions, then we interleave to break the cross // iteration dependency. // 2. If the loop is really small, then we interleave to reduce the loop // overhead. // 3. We don't interleave if we think that we will spill registers to memory // due to the increased register pressure. if (!isScalarEpilogueAllowed()) return 1; // We used the distance for the interleave count. if (!Legal->isSafeForAnyVectorWidth()) return 1; auto BestKnownTC = getSmallBestKnownTC(*PSE.getSE(), TheLoop); const bool HasReductions = !Legal->getReductionVars().empty(); // Do not interleave loops with a relatively small known or estimated trip // count. But we will interleave when InterleaveSmallLoopScalarReduction is // enabled, and the code has scalar reductions(HasReductions && VF = 1), // because with the above conditions interleaving can expose ILP and break // cross iteration dependences for reductions. if (BestKnownTC && (*BestKnownTC < TinyTripCountInterleaveThreshold) && !(InterleaveSmallLoopScalarReduction && HasReductions && VF.isScalar())) return 1; // If we did not calculate the cost for VF (because the user selected the VF) // then we calculate the cost of VF here. if (LoopCost == 0) { LoopCost = expectedCost(VF).first; assert(LoopCost.isValid() && "Expected to have chosen a VF with valid cost"); // Loop body is free and there is no need for interleaving. if (LoopCost == 0) return 1; } RegisterUsage R = calculateRegisterUsage({VF})[0]; // We divide by these constants so assume that we have at least one // instruction that uses at least one register. for (auto& pair : R.MaxLocalUsers) { pair.second = std::max(pair.second, 1U); } // We calculate the interleave count using the following formula. // Subtract the number of loop invariants from the number of available // registers. These registers are used by all of the interleaved instances. // Next, divide the remaining registers by the number of registers that is // required by the loop, in order to estimate how many parallel instances // fit without causing spills. All of this is rounded down if necessary to be // a power of two. We want power of two interleave count to simplify any // addressing operations or alignment considerations. // We also want power of two interleave counts to ensure that the induction // variable of the vector loop wraps to zero, when tail is folded by masking; // this currently happens when OptForSize, in which case IC is set to 1 above. unsigned IC = UINT_MAX; for (auto& pair : R.MaxLocalUsers) { unsigned TargetNumRegisters = TTI.getNumberOfRegisters(pair.first); LLVM_DEBUG(dbgs() << "LV: The target has " << TargetNumRegisters << " registers of " << TTI.getRegisterClassName(pair.first) << " register class\n"); if (VF.isScalar()) { if (ForceTargetNumScalarRegs.getNumOccurrences() > 0) TargetNumRegisters = ForceTargetNumScalarRegs; } else { if (ForceTargetNumVectorRegs.getNumOccurrences() > 0) TargetNumRegisters = ForceTargetNumVectorRegs; } unsigned MaxLocalUsers = pair.second; unsigned LoopInvariantRegs = 0; if (R.LoopInvariantRegs.find(pair.first) != R.LoopInvariantRegs.end()) LoopInvariantRegs = R.LoopInvariantRegs[pair.first]; unsigned TmpIC = llvm::bit_floor((TargetNumRegisters - LoopInvariantRegs) / MaxLocalUsers); // Don't count the induction variable as interleaved. if (EnableIndVarRegisterHeur) { TmpIC = llvm::bit_floor((TargetNumRegisters - LoopInvariantRegs - 1) / std::max(1U, (MaxLocalUsers - 1))); } IC = std::min(IC, TmpIC); } // Clamp the interleave ranges to reasonable counts. unsigned MaxInterleaveCount = TTI.getMaxInterleaveFactor(VF); // Check if the user has overridden the max. if (VF.isScalar()) { if (ForceTargetMaxScalarInterleaveFactor.getNumOccurrences() > 0) MaxInterleaveCount = ForceTargetMaxScalarInterleaveFactor; } else { if (ForceTargetMaxVectorInterleaveFactor.getNumOccurrences() > 0) MaxInterleaveCount = ForceTargetMaxVectorInterleaveFactor; } unsigned EstimatedVF = VF.getKnownMinValue(); if (VF.isScalable()) { if (std::optional VScale = getVScaleForTuning(TheLoop, TTI)) EstimatedVF *= *VScale; } assert(EstimatedVF >= 1 && "Estimated VF shouldn't be less than 1"); unsigned KnownTC = PSE.getSE()->getSmallConstantTripCount(TheLoop); if (KnownTC) { // If trip count is known we select between two prospective ICs, where // 1) the aggressive IC is capped by the trip count divided by VF // 2) the conservative IC is capped by the trip count divided by (VF * 2) // The final IC is selected in a way that the epilogue loop trip count is // minimized while maximizing the IC itself, so that we either run the // vector loop at least once if it generates a small epilogue loop, or else // we run the vector loop at least twice. unsigned InterleaveCountUB = bit_floor( std::max(1u, std::min(KnownTC / EstimatedVF, MaxInterleaveCount))); unsigned InterleaveCountLB = bit_floor(std::max( 1u, std::min(KnownTC / (EstimatedVF * 2), MaxInterleaveCount))); MaxInterleaveCount = InterleaveCountLB; if (InterleaveCountUB != InterleaveCountLB) { unsigned TailTripCountUB = (KnownTC % (EstimatedVF * InterleaveCountUB)); unsigned TailTripCountLB = (KnownTC % (EstimatedVF * InterleaveCountLB)); // If both produce same scalar tail, maximize the IC to do the same work // in fewer vector loop iterations if (TailTripCountUB == TailTripCountLB) MaxInterleaveCount = InterleaveCountUB; } } else if (BestKnownTC) { // If trip count is an estimated compile time constant, limit the // IC to be capped by the trip count divided by VF * 2, such that the vector // loop runs at least twice to make interleaving seem profitable when there // is an epilogue loop present. Since exact Trip count is not known we // choose to be conservative in our IC estimate. MaxInterleaveCount = bit_floor(std::max( 1u, std::min(*BestKnownTC / (EstimatedVF * 2), MaxInterleaveCount))); } assert(MaxInterleaveCount > 0 && "Maximum interleave count must be greater than 0"); // Clamp the calculated IC to be between the 1 and the max interleave count // that the target and trip count allows. if (IC > MaxInterleaveCount) IC = MaxInterleaveCount; else // Make sure IC is greater than 0. IC = std::max(1u, IC); assert(IC > 0 && "Interleave count must be greater than 0."); // Interleave if we vectorized this loop and there is a reduction that could // benefit from interleaving. if (VF.isVector() && HasReductions) { LLVM_DEBUG(dbgs() << "LV: Interleaving because of reductions.\n"); return IC; } // For any scalar loop that either requires runtime checks or predication we // are better off leaving this to the unroller. Note that if we've already // vectorized the loop we will have done the runtime check and so interleaving // won't require further checks. bool ScalarInterleavingRequiresPredication = (VF.isScalar() && any_of(TheLoop->blocks(), [this](BasicBlock *BB) { return Legal->blockNeedsPredication(BB); })); bool ScalarInterleavingRequiresRuntimePointerCheck = (VF.isScalar() && Legal->getRuntimePointerChecking()->Need); // We want to interleave small loops in order to reduce the loop overhead and // potentially expose ILP opportunities. LLVM_DEBUG(dbgs() << "LV: Loop cost is " << LoopCost << '\n' << "LV: IC is " << IC << '\n' << "LV: VF is " << VF << '\n'); const bool AggressivelyInterleaveReductions = TTI.enableAggressiveInterleaving(HasReductions); if (!ScalarInterleavingRequiresRuntimePointerCheck && !ScalarInterleavingRequiresPredication && LoopCost < SmallLoopCost) { // We assume that the cost overhead is 1 and we use the cost model // to estimate the cost of the loop and interleave until the cost of the // loop overhead is about 5% of the cost of the loop. unsigned SmallIC = std::min(IC, (unsigned)llvm::bit_floor( SmallLoopCost / *LoopCost.getValue())); // Interleave until store/load ports (estimated by max interleave count) are // saturated. unsigned NumStores = Legal->getNumStores(); unsigned NumLoads = Legal->getNumLoads(); unsigned StoresIC = IC / (NumStores ? NumStores : 1); unsigned LoadsIC = IC / (NumLoads ? NumLoads : 1); // There is little point in interleaving for reductions containing selects // and compares when VF=1 since it may just create more overhead than it's // worth for loops with small trip counts. This is because we still have to // do the final reduction after the loop. bool HasSelectCmpReductions = HasReductions && any_of(Legal->getReductionVars(), [&](auto &Reduction) -> bool { const RecurrenceDescriptor &RdxDesc = Reduction.second; return RecurrenceDescriptor::isAnyOfRecurrenceKind( RdxDesc.getRecurrenceKind()); }); if (HasSelectCmpReductions) { LLVM_DEBUG(dbgs() << "LV: Not interleaving select-cmp reductions.\n"); return 1; } // If we have a scalar reduction (vector reductions are already dealt with // by this point), we can increase the critical path length if the loop // we're interleaving is inside another loop. For tree-wise reductions // set the limit to 2, and for ordered reductions it's best to disable // interleaving entirely. if (HasReductions && TheLoop->getLoopDepth() > 1) { bool HasOrderedReductions = any_of(Legal->getReductionVars(), [&](auto &Reduction) -> bool { const RecurrenceDescriptor &RdxDesc = Reduction.second; return RdxDesc.isOrdered(); }); if (HasOrderedReductions) { LLVM_DEBUG( dbgs() << "LV: Not interleaving scalar ordered reductions.\n"); return 1; } unsigned F = static_cast(MaxNestedScalarReductionIC); SmallIC = std::min(SmallIC, F); StoresIC = std::min(StoresIC, F); LoadsIC = std::min(LoadsIC, F); } if (EnableLoadStoreRuntimeInterleave && std::max(StoresIC, LoadsIC) > SmallIC) { LLVM_DEBUG( dbgs() << "LV: Interleaving to saturate store or load ports.\n"); return std::max(StoresIC, LoadsIC); } // If there are scalar reductions and TTI has enabled aggressive // interleaving for reductions, we will interleave to expose ILP. if (InterleaveSmallLoopScalarReduction && VF.isScalar() && AggressivelyInterleaveReductions) { LLVM_DEBUG(dbgs() << "LV: Interleaving to expose ILP.\n"); // Interleave no less than SmallIC but not as aggressive as the normal IC // to satisfy the rare situation when resources are too limited. return std::max(IC / 2, SmallIC); } else { LLVM_DEBUG(dbgs() << "LV: Interleaving to reduce branch cost.\n"); return SmallIC; } } // Interleave if this is a large loop (small loops are already dealt with by // this point) that could benefit from interleaving. if (AggressivelyInterleaveReductions) { LLVM_DEBUG(dbgs() << "LV: Interleaving to expose ILP.\n"); return IC; } LLVM_DEBUG(dbgs() << "LV: Not Interleaving.\n"); return 1; } SmallVector LoopVectorizationCostModel::calculateRegisterUsage(ArrayRef VFs) { // This function calculates the register usage by measuring the highest number // of values that are alive at a single location. Obviously, this is a very // rough estimation. We scan the loop in a topological order in order and // assign a number to each instruction. We use RPO to ensure that defs are // met before their users. We assume that each instruction that has in-loop // users starts an interval. We record every time that an in-loop value is // used, so we have a list of the first and last occurrences of each // instruction. Next, we transpose this data structure into a multi map that // holds the list of intervals that *end* at a specific location. This multi // map allows us to perform a linear search. We scan the instructions linearly // and record each time that a new interval starts, by placing it in a set. // If we find this value in the multi-map then we remove it from the set. // The max register usage is the maximum size of the set. // We also search for instructions that are defined outside the loop, but are // used inside the loop. We need this number separately from the max-interval // usage number because when we unroll, loop-invariant values do not take // more register. LoopBlocksDFS DFS(TheLoop); DFS.perform(LI); RegisterUsage RU; // Each 'key' in the map opens a new interval. The values // of the map are the index of the 'last seen' usage of the // instruction that is the key. using IntervalMap = DenseMap; // Maps instruction to its index. SmallVector IdxToInstr; // Marks the end of each interval. IntervalMap EndPoint; // Saves the list of instruction indices that are used in the loop. SmallPtrSet Ends; // Saves the list of values that are used in the loop but are defined outside // the loop (not including non-instruction values such as arguments and // constants). SmallSetVector LoopInvariants; for (BasicBlock *BB : make_range(DFS.beginRPO(), DFS.endRPO())) { for (Instruction &I : BB->instructionsWithoutDebug()) { IdxToInstr.push_back(&I); // Save the end location of each USE. for (Value *U : I.operands()) { auto *Instr = dyn_cast(U); // Ignore non-instruction values such as arguments, constants, etc. // FIXME: Might need some motivation why these values are ignored. If // for example an argument is used inside the loop it will increase the // register pressure (so shouldn't we add it to LoopInvariants). if (!Instr) continue; // If this instruction is outside the loop then record it and continue. if (!TheLoop->contains(Instr)) { LoopInvariants.insert(Instr); continue; } // Overwrite previous end points. EndPoint[Instr] = IdxToInstr.size(); Ends.insert(Instr); } } } // Saves the list of intervals that end with the index in 'key'. using InstrList = SmallVector; DenseMap TransposeEnds; // Transpose the EndPoints to a list of values that end at each index. for (auto &Interval : EndPoint) TransposeEnds[Interval.second].push_back(Interval.first); SmallPtrSet OpenIntervals; SmallVector RUs(VFs.size()); SmallVector, 8> MaxUsages(VFs.size()); LLVM_DEBUG(dbgs() << "LV(REG): Calculating max register usage:\n"); const auto &TTICapture = TTI; auto GetRegUsage = [&TTICapture](Type *Ty, ElementCount VF) -> unsigned { if (Ty->isTokenTy() || !VectorType::isValidElementType(Ty)) return 0; return TTICapture.getRegUsageForType(VectorType::get(Ty, VF)); }; for (unsigned int i = 0, s = IdxToInstr.size(); i < s; ++i) { Instruction *I = IdxToInstr[i]; // Remove all of the instructions that end at this location. InstrList &List = TransposeEnds[i]; for (Instruction *ToRemove : List) OpenIntervals.erase(ToRemove); // Ignore instructions that are never used within the loop. if (!Ends.count(I)) continue; // Skip ignored values. if (ValuesToIgnore.count(I)) continue; collectInLoopReductions(); // For each VF find the maximum usage of registers. for (unsigned j = 0, e = VFs.size(); j < e; ++j) { // Count the number of registers used, per register class, given all open // intervals. // Note that elements in this SmallMapVector will be default constructed // as 0. So we can use "RegUsage[ClassID] += n" in the code below even if // there is no previous entry for ClassID. SmallMapVector RegUsage; if (VFs[j].isScalar()) { for (auto *Inst : OpenIntervals) { unsigned ClassID = TTI.getRegisterClassForType(false, Inst->getType()); // FIXME: The target might use more than one register for the type // even in the scalar case. RegUsage[ClassID] += 1; } } else { collectUniformsAndScalars(VFs[j]); for (auto *Inst : OpenIntervals) { // Skip ignored values for VF > 1. if (VecValuesToIgnore.count(Inst)) continue; if (isScalarAfterVectorization(Inst, VFs[j])) { unsigned ClassID = TTI.getRegisterClassForType(false, Inst->getType()); // FIXME: The target might use more than one register for the type // even in the scalar case. RegUsage[ClassID] += 1; } else { unsigned ClassID = TTI.getRegisterClassForType(true, Inst->getType()); RegUsage[ClassID] += GetRegUsage(Inst->getType(), VFs[j]); } } } for (auto& pair : RegUsage) { auto &Entry = MaxUsages[j][pair.first]; Entry = std::max(Entry, pair.second); } } LLVM_DEBUG(dbgs() << "LV(REG): At #" << i << " Interval # " << OpenIntervals.size() << '\n'); // Add the current instruction to the list of open intervals. OpenIntervals.insert(I); } for (unsigned i = 0, e = VFs.size(); i < e; ++i) { // Note that elements in this SmallMapVector will be default constructed // as 0. So we can use "Invariant[ClassID] += n" in the code below even if // there is no previous entry for ClassID. SmallMapVector Invariant; for (auto *Inst : LoopInvariants) { // FIXME: The target might use more than one register for the type // even in the scalar case. bool IsScalar = all_of(Inst->users(), [&](User *U) { auto *I = cast(U); return TheLoop != LI->getLoopFor(I->getParent()) || isScalarAfterVectorization(I, VFs[i]); }); ElementCount VF = IsScalar ? ElementCount::getFixed(1) : VFs[i]; unsigned ClassID = TTI.getRegisterClassForType(VF.isVector(), Inst->getType()); Invariant[ClassID] += GetRegUsage(Inst->getType(), VF); } LLVM_DEBUG({ dbgs() << "LV(REG): VF = " << VFs[i] << '\n'; dbgs() << "LV(REG): Found max usage: " << MaxUsages[i].size() << " item\n"; for (const auto &pair : MaxUsages[i]) { dbgs() << "LV(REG): RegisterClass: " << TTI.getRegisterClassName(pair.first) << ", " << pair.second << " registers\n"; } dbgs() << "LV(REG): Found invariant usage: " << Invariant.size() << " item\n"; for (const auto &pair : Invariant) { dbgs() << "LV(REG): RegisterClass: " << TTI.getRegisterClassName(pair.first) << ", " << pair.second << " registers\n"; } }); RU.LoopInvariantRegs = Invariant; RU.MaxLocalUsers = MaxUsages[i]; RUs[i] = RU; } return RUs; } bool LoopVectorizationCostModel::useEmulatedMaskMemRefHack(Instruction *I, ElementCount VF) { // TODO: Cost model for emulated masked load/store is completely // broken. This hack guides the cost model to use an artificially // high enough value to practically disable vectorization with such // operations, except where previously deployed legality hack allowed // using very low cost values. This is to avoid regressions coming simply // from moving "masked load/store" check from legality to cost model. // Masked Load/Gather emulation was previously never allowed. // Limited number of Masked Store/Scatter emulation was allowed. assert((isPredicatedInst(I)) && "Expecting a scalar emulated instruction"); return isa(I) || (isa(I) && NumPredStores > NumberOfStoresToPredicate); } void LoopVectorizationCostModel::collectInstsToScalarize(ElementCount VF) { // If we aren't vectorizing the loop, or if we've already collected the // instructions to scalarize, there's nothing to do. Collection may already // have occurred if we have a user-selected VF and are now computing the // expected cost for interleaving. if (VF.isScalar() || VF.isZero() || InstsToScalarize.contains(VF)) return; // Initialize a mapping for VF in InstsToScalalarize. If we find that it's // not profitable to scalarize any instructions, the presence of VF in the // map will indicate that we've analyzed it already. ScalarCostsTy &ScalarCostsVF = InstsToScalarize[VF]; PredicatedBBsAfterVectorization[VF].clear(); // Find all the instructions that are scalar with predication in the loop and // determine if it would be better to not if-convert the blocks they are in. // If so, we also record the instructions to scalarize. for (BasicBlock *BB : TheLoop->blocks()) { if (!blockNeedsPredicationForAnyReason(BB)) continue; for (Instruction &I : *BB) if (isScalarWithPredication(&I, VF)) { ScalarCostsTy ScalarCosts; // Do not apply discount if scalable, because that would lead to // invalid scalarization costs. // Do not apply discount logic if hacked cost is needed // for emulated masked memrefs. if (!VF.isScalable() && !useEmulatedMaskMemRefHack(&I, VF) && computePredInstDiscount(&I, ScalarCosts, VF) >= 0) ScalarCostsVF.insert(ScalarCosts.begin(), ScalarCosts.end()); // Remember that BB will remain after vectorization. PredicatedBBsAfterVectorization[VF].insert(BB); } } } InstructionCost LoopVectorizationCostModel::computePredInstDiscount( Instruction *PredInst, ScalarCostsTy &ScalarCosts, ElementCount VF) { assert(!isUniformAfterVectorization(PredInst, VF) && "Instruction marked uniform-after-vectorization will be predicated"); // Initialize the discount to zero, meaning that the scalar version and the // vector version cost the same. InstructionCost Discount = 0; // Holds instructions to analyze. The instructions we visit are mapped in // ScalarCosts. Those instructions are the ones that would be scalarized if // we find that the scalar version costs less. SmallVector Worklist; // Returns true if the given instruction can be scalarized. auto canBeScalarized = [&](Instruction *I) -> bool { // We only attempt to scalarize instructions forming a single-use chain // from the original predicated block that would otherwise be vectorized. // Although not strictly necessary, we give up on instructions we know will // already be scalar to avoid traversing chains that are unlikely to be // beneficial. if (!I->hasOneUse() || PredInst->getParent() != I->getParent() || isScalarAfterVectorization(I, VF)) return false; // If the instruction is scalar with predication, it will be analyzed // separately. We ignore it within the context of PredInst. if (isScalarWithPredication(I, VF)) return false; // If any of the instruction's operands are uniform after vectorization, // the instruction cannot be scalarized. This prevents, for example, a // masked load from being scalarized. // // We assume we will only emit a value for lane zero of an instruction // marked uniform after vectorization, rather than VF identical values. // Thus, if we scalarize an instruction that uses a uniform, we would // create uses of values corresponding to the lanes we aren't emitting code // for. This behavior can be changed by allowing getScalarValue to clone // the lane zero values for uniforms rather than asserting. for (Use &U : I->operands()) if (auto *J = dyn_cast(U.get())) if (isUniformAfterVectorization(J, VF)) return false; // Otherwise, we can scalarize the instruction. return true; }; // Compute the expected cost discount from scalarizing the entire expression // feeding the predicated instruction. We currently only consider expressions // that are single-use instruction chains. Worklist.push_back(PredInst); while (!Worklist.empty()) { Instruction *I = Worklist.pop_back_val(); // If we've already analyzed the instruction, there's nothing to do. if (ScalarCosts.contains(I)) continue; // Compute the cost of the vector instruction. Note that this cost already // includes the scalarization overhead of the predicated instruction. InstructionCost VectorCost = getInstructionCost(I, VF).first; // Compute the cost of the scalarized instruction. This cost is the cost of // the instruction as if it wasn't if-converted and instead remained in the // predicated block. We will scale this cost by block probability after // computing the scalarization overhead. InstructionCost ScalarCost = VF.getFixedValue() * getInstructionCost(I, ElementCount::getFixed(1)).first; // Compute the scalarization overhead of needed insertelement instructions // and phi nodes. TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput; if (isScalarWithPredication(I, VF) && !I->getType()->isVoidTy()) { ScalarCost += TTI.getScalarizationOverhead( cast(ToVectorTy(I->getType(), VF)), APInt::getAllOnes(VF.getFixedValue()), /*Insert*/ true, /*Extract*/ false, CostKind); ScalarCost += VF.getFixedValue() * TTI.getCFInstrCost(Instruction::PHI, CostKind); } // Compute the scalarization overhead of needed extractelement // instructions. For each of the instruction's operands, if the operand can // be scalarized, add it to the worklist; otherwise, account for the // overhead. for (Use &U : I->operands()) if (auto *J = dyn_cast(U.get())) { assert(VectorType::isValidElementType(J->getType()) && "Instruction has non-scalar type"); if (canBeScalarized(J)) Worklist.push_back(J); else if (needsExtract(J, VF)) { ScalarCost += TTI.getScalarizationOverhead( cast(ToVectorTy(J->getType(), VF)), APInt::getAllOnes(VF.getFixedValue()), /*Insert*/ false, /*Extract*/ true, CostKind); } } // Scale the total scalar cost by block probability. ScalarCost /= getReciprocalPredBlockProb(); // Compute the discount. A non-negative discount means the vector version // of the instruction costs more, and scalarizing would be beneficial. Discount += VectorCost - ScalarCost; ScalarCosts[I] = ScalarCost; } return Discount; } LoopVectorizationCostModel::VectorizationCostTy LoopVectorizationCostModel::expectedCost( ElementCount VF, SmallVectorImpl *Invalid) { VectorizationCostTy Cost; // For each block. for (BasicBlock *BB : TheLoop->blocks()) { VectorizationCostTy BlockCost; // For each instruction in the old loop. for (Instruction &I : BB->instructionsWithoutDebug()) { // Skip ignored values. if (ValuesToIgnore.count(&I) || (VF.isVector() && VecValuesToIgnore.count(&I))) continue; VectorizationCostTy C = getInstructionCost(&I, VF); // Check if we should override the cost. if (C.first.isValid() && ForceTargetInstructionCost.getNumOccurrences() > 0) C.first = InstructionCost(ForceTargetInstructionCost); // Keep a list of instructions with invalid costs. if (Invalid && !C.first.isValid()) Invalid->emplace_back(&I, VF); BlockCost.first += C.first; BlockCost.second |= C.second; LLVM_DEBUG(dbgs() << "LV: Found an estimated cost of " << C.first << " for VF " << VF << " For instruction: " << I << '\n'); } // If we are vectorizing a predicated block, it will have been // if-converted. This means that the block's instructions (aside from // stores and instructions that may divide by zero) will now be // unconditionally executed. For the scalar case, we may not always execute // the predicated block, if it is an if-else block. Thus, scale the block's // cost by the probability of executing it. blockNeedsPredication from // Legal is used so as to not include all blocks in tail folded loops. if (VF.isScalar() && Legal->blockNeedsPredication(BB)) BlockCost.first /= getReciprocalPredBlockProb(); Cost.first += BlockCost.first; Cost.second |= BlockCost.second; } return Cost; } /// Gets Address Access SCEV after verifying that the access pattern /// is loop invariant except the induction variable dependence. /// /// This SCEV can be sent to the Target in order to estimate the address /// calculation cost. static const SCEV *getAddressAccessSCEV( Value *Ptr, LoopVectorizationLegality *Legal, PredicatedScalarEvolution &PSE, const Loop *TheLoop) { auto *Gep = dyn_cast(Ptr); if (!Gep) return nullptr; // We are looking for a gep with all loop invariant indices except for one // which should be an induction variable. auto SE = PSE.getSE(); unsigned NumOperands = Gep->getNumOperands(); for (unsigned i = 1; i < NumOperands; ++i) { Value *Opd = Gep->getOperand(i); if (!SE->isLoopInvariant(SE->getSCEV(Opd), TheLoop) && !Legal->isInductionVariable(Opd)) return nullptr; } // Now we know we have a GEP ptr, %inv, %ind, %inv. return the Ptr SCEV. return PSE.getSCEV(Ptr); } InstructionCost LoopVectorizationCostModel::getMemInstScalarizationCost(Instruction *I, ElementCount VF) { assert(VF.isVector() && "Scalarization cost of instruction implies vectorization."); if (VF.isScalable()) return InstructionCost::getInvalid(); Type *ValTy = getLoadStoreType(I); auto SE = PSE.getSE(); unsigned AS = getLoadStoreAddressSpace(I); Value *Ptr = getLoadStorePointerOperand(I); Type *PtrTy = ToVectorTy(Ptr->getType(), VF); // NOTE: PtrTy is a vector to signal `TTI::getAddressComputationCost` // that it is being called from this specific place. // Figure out whether the access is strided and get the stride value // if it's known in compile time const SCEV *PtrSCEV = getAddressAccessSCEV(Ptr, Legal, PSE, TheLoop); // Get the cost of the scalar memory instruction and address computation. InstructionCost Cost = VF.getKnownMinValue() * TTI.getAddressComputationCost(PtrTy, SE, PtrSCEV); // Don't pass *I here, since it is scalar but will actually be part of a // vectorized loop where the user of it is a vectorized instruction. TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput; const Align Alignment = getLoadStoreAlignment(I); Cost += VF.getKnownMinValue() * TTI.getMemoryOpCost(I->getOpcode(), ValTy->getScalarType(), Alignment, AS, CostKind); // Get the overhead of the extractelement and insertelement instructions // we might create due to scalarization. Cost += getScalarizationOverhead(I, VF, CostKind); // If we have a predicated load/store, it will need extra i1 extracts and // conditional branches, but may not be executed for each vector lane. Scale // the cost by the probability of executing the predicated block. if (isPredicatedInst(I)) { Cost /= getReciprocalPredBlockProb(); // Add the cost of an i1 extract and a branch auto *Vec_i1Ty = VectorType::get(IntegerType::getInt1Ty(ValTy->getContext()), VF); Cost += TTI.getScalarizationOverhead( Vec_i1Ty, APInt::getAllOnes(VF.getKnownMinValue()), /*Insert=*/false, /*Extract=*/true, CostKind); Cost += TTI.getCFInstrCost(Instruction::Br, CostKind); if (useEmulatedMaskMemRefHack(I, VF)) // Artificially setting to a high enough value to practically disable // vectorization with such operations. Cost = 3000000; } return Cost; } InstructionCost LoopVectorizationCostModel::getConsecutiveMemOpCost(Instruction *I, ElementCount VF) { Type *ValTy = getLoadStoreType(I); auto *VectorTy = cast(ToVectorTy(ValTy, VF)); Value *Ptr = getLoadStorePointerOperand(I); unsigned AS = getLoadStoreAddressSpace(I); int ConsecutiveStride = Legal->isConsecutivePtr(ValTy, Ptr); enum TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput; assert((ConsecutiveStride == 1 || ConsecutiveStride == -1) && "Stride should be 1 or -1 for consecutive memory access"); const Align Alignment = getLoadStoreAlignment(I); InstructionCost Cost = 0; if (Legal->isMaskRequired(I)) { Cost += TTI.getMaskedMemoryOpCost(I->getOpcode(), VectorTy, Alignment, AS, CostKind); } else { TTI::OperandValueInfo OpInfo = TTI::getOperandInfo(I->getOperand(0)); Cost += TTI.getMemoryOpCost(I->getOpcode(), VectorTy, Alignment, AS, CostKind, OpInfo, I); } bool Reverse = ConsecutiveStride < 0; if (Reverse) Cost += TTI.getShuffleCost(TargetTransformInfo::SK_Reverse, VectorTy, std::nullopt, CostKind, 0); return Cost; } InstructionCost LoopVectorizationCostModel::getUniformMemOpCost(Instruction *I, ElementCount VF) { assert(Legal->isUniformMemOp(*I, VF)); Type *ValTy = getLoadStoreType(I); auto *VectorTy = cast(ToVectorTy(ValTy, VF)); const Align Alignment = getLoadStoreAlignment(I); unsigned AS = getLoadStoreAddressSpace(I); enum TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput; if (isa(I)) { return TTI.getAddressComputationCost(ValTy) + TTI.getMemoryOpCost(Instruction::Load, ValTy, Alignment, AS, CostKind) + TTI.getShuffleCost(TargetTransformInfo::SK_Broadcast, VectorTy); } StoreInst *SI = cast(I); bool isLoopInvariantStoreValue = Legal->isInvariant(SI->getValueOperand()); return TTI.getAddressComputationCost(ValTy) + TTI.getMemoryOpCost(Instruction::Store, ValTy, Alignment, AS, CostKind) + (isLoopInvariantStoreValue ? 0 : TTI.getVectorInstrCost(Instruction::ExtractElement, VectorTy, CostKind, VF.getKnownMinValue() - 1)); } InstructionCost LoopVectorizationCostModel::getGatherScatterCost(Instruction *I, ElementCount VF) { Type *ValTy = getLoadStoreType(I); auto *VectorTy = cast(ToVectorTy(ValTy, VF)); const Align Alignment = getLoadStoreAlignment(I); const Value *Ptr = getLoadStorePointerOperand(I); return TTI.getAddressComputationCost(VectorTy) + TTI.getGatherScatterOpCost( I->getOpcode(), VectorTy, Ptr, Legal->isMaskRequired(I), Alignment, TargetTransformInfo::TCK_RecipThroughput, I); } InstructionCost LoopVectorizationCostModel::getInterleaveGroupCost(Instruction *I, ElementCount VF) { Type *ValTy = getLoadStoreType(I); auto *VectorTy = cast(ToVectorTy(ValTy, VF)); unsigned AS = getLoadStoreAddressSpace(I); enum TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput; auto Group = getInterleavedAccessGroup(I); assert(Group && "Fail to get an interleaved access group."); unsigned InterleaveFactor = Group->getFactor(); auto *WideVecTy = VectorType::get(ValTy, VF * InterleaveFactor); // Holds the indices of existing members in the interleaved group. SmallVector Indices; for (unsigned IF = 0; IF < InterleaveFactor; IF++) if (Group->getMember(IF)) Indices.push_back(IF); // Calculate the cost of the whole interleaved group. bool UseMaskForGaps = (Group->requiresScalarEpilogue() && !isScalarEpilogueAllowed()) || (isa(I) && (Group->getNumMembers() < Group->getFactor())); InstructionCost Cost = TTI.getInterleavedMemoryOpCost( I->getOpcode(), WideVecTy, Group->getFactor(), Indices, Group->getAlign(), AS, CostKind, Legal->isMaskRequired(I), UseMaskForGaps); if (Group->isReverse()) { // TODO: Add support for reversed masked interleaved access. assert(!Legal->isMaskRequired(I) && "Reverse masked interleaved access not supported."); Cost += Group->getNumMembers() * TTI.getShuffleCost(TargetTransformInfo::SK_Reverse, VectorTy, std::nullopt, CostKind, 0); } return Cost; } std::optional LoopVectorizationCostModel::getReductionPatternCost( Instruction *I, ElementCount VF, Type *Ty, TTI::TargetCostKind CostKind) const { using namespace llvm::PatternMatch; // Early exit for no inloop reductions if (InLoopReductions.empty() || VF.isScalar() || !isa(Ty)) return std::nullopt; auto *VectorTy = cast(Ty); // We are looking for a pattern of, and finding the minimal acceptable cost: // reduce(mul(ext(A), ext(B))) or // reduce(mul(A, B)) or // reduce(ext(A)) or // reduce(A). // The basic idea is that we walk down the tree to do that, finding the root // reduction instruction in InLoopReductionImmediateChains. From there we find // the pattern of mul/ext and test the cost of the entire pattern vs the cost // of the components. If the reduction cost is lower then we return it for the // reduction instruction and 0 for the other instructions in the pattern. If // it is not we return an invalid cost specifying the orignal cost method // should be used. Instruction *RetI = I; if (match(RetI, m_ZExtOrSExt(m_Value()))) { if (!RetI->hasOneUser()) return std::nullopt; RetI = RetI->user_back(); } if (match(RetI, m_OneUse(m_Mul(m_Value(), m_Value()))) && RetI->user_back()->getOpcode() == Instruction::Add) { RetI = RetI->user_back(); } // Test if the found instruction is a reduction, and if not return an invalid // cost specifying the parent to use the original cost modelling. if (!InLoopReductionImmediateChains.count(RetI)) return std::nullopt; // Find the reduction this chain is a part of and calculate the basic cost of // the reduction on its own. Instruction *LastChain = InLoopReductionImmediateChains.at(RetI); Instruction *ReductionPhi = LastChain; while (!isa(ReductionPhi)) ReductionPhi = InLoopReductionImmediateChains.at(ReductionPhi); const RecurrenceDescriptor &RdxDesc = Legal->getReductionVars().find(cast(ReductionPhi))->second; InstructionCost BaseCost = TTI.getArithmeticReductionCost( RdxDesc.getOpcode(), VectorTy, RdxDesc.getFastMathFlags(), CostKind); // For a call to the llvm.fmuladd intrinsic we need to add the cost of a // normal fmul instruction to the cost of the fadd reduction. if (RdxDesc.getRecurrenceKind() == RecurKind::FMulAdd) BaseCost += TTI.getArithmeticInstrCost(Instruction::FMul, VectorTy, CostKind); // If we're using ordered reductions then we can just return the base cost // here, since getArithmeticReductionCost calculates the full ordered // reduction cost when FP reassociation is not allowed. if (useOrderedReductions(RdxDesc)) return BaseCost; // Get the operand that was not the reduction chain and match it to one of the // patterns, returning the better cost if it is found. Instruction *RedOp = RetI->getOperand(1) == LastChain ? dyn_cast(RetI->getOperand(0)) : dyn_cast(RetI->getOperand(1)); VectorTy = VectorType::get(I->getOperand(0)->getType(), VectorTy); Instruction *Op0, *Op1; if (RedOp && RdxDesc.getOpcode() == Instruction::Add && match(RedOp, m_ZExtOrSExt(m_Mul(m_Instruction(Op0), m_Instruction(Op1)))) && match(Op0, m_ZExtOrSExt(m_Value())) && Op0->getOpcode() == Op1->getOpcode() && Op0->getOperand(0)->getType() == Op1->getOperand(0)->getType() && !TheLoop->isLoopInvariant(Op0) && !TheLoop->isLoopInvariant(Op1) && (Op0->getOpcode() == RedOp->getOpcode() || Op0 == Op1)) { // Matched reduce.add(ext(mul(ext(A), ext(B))) // Note that the extend opcodes need to all match, or if A==B they will have // been converted to zext(mul(sext(A), sext(A))) as it is known positive, // which is equally fine. bool IsUnsigned = isa(Op0); auto *ExtType = VectorType::get(Op0->getOperand(0)->getType(), VectorTy); auto *MulType = VectorType::get(Op0->getType(), VectorTy); InstructionCost ExtCost = TTI.getCastInstrCost(Op0->getOpcode(), MulType, ExtType, TTI::CastContextHint::None, CostKind, Op0); InstructionCost MulCost = TTI.getArithmeticInstrCost(Instruction::Mul, MulType, CostKind); InstructionCost Ext2Cost = TTI.getCastInstrCost(RedOp->getOpcode(), VectorTy, MulType, TTI::CastContextHint::None, CostKind, RedOp); InstructionCost RedCost = TTI.getMulAccReductionCost( IsUnsigned, RdxDesc.getRecurrenceType(), ExtType, CostKind); if (RedCost.isValid() && RedCost < ExtCost * 2 + MulCost + Ext2Cost + BaseCost) return I == RetI ? RedCost : 0; } else if (RedOp && match(RedOp, m_ZExtOrSExt(m_Value())) && !TheLoop->isLoopInvariant(RedOp)) { // Matched reduce(ext(A)) bool IsUnsigned = isa(RedOp); auto *ExtType = VectorType::get(RedOp->getOperand(0)->getType(), VectorTy); InstructionCost RedCost = TTI.getExtendedReductionCost( RdxDesc.getOpcode(), IsUnsigned, RdxDesc.getRecurrenceType(), ExtType, RdxDesc.getFastMathFlags(), CostKind); InstructionCost ExtCost = TTI.getCastInstrCost(RedOp->getOpcode(), VectorTy, ExtType, TTI::CastContextHint::None, CostKind, RedOp); if (RedCost.isValid() && RedCost < BaseCost + ExtCost) return I == RetI ? RedCost : 0; } else if (RedOp && RdxDesc.getOpcode() == Instruction::Add && match(RedOp, m_Mul(m_Instruction(Op0), m_Instruction(Op1)))) { if (match(Op0, m_ZExtOrSExt(m_Value())) && Op0->getOpcode() == Op1->getOpcode() && !TheLoop->isLoopInvariant(Op0) && !TheLoop->isLoopInvariant(Op1)) { bool IsUnsigned = isa(Op0); Type *Op0Ty = Op0->getOperand(0)->getType(); Type *Op1Ty = Op1->getOperand(0)->getType(); Type *LargestOpTy = Op0Ty->getIntegerBitWidth() < Op1Ty->getIntegerBitWidth() ? Op1Ty : Op0Ty; auto *ExtType = VectorType::get(LargestOpTy, VectorTy); // Matched reduce.add(mul(ext(A), ext(B))), where the two ext may be of // different sizes. We take the largest type as the ext to reduce, and add // the remaining cost as, for example reduce(mul(ext(ext(A)), ext(B))). InstructionCost ExtCost0 = TTI.getCastInstrCost( Op0->getOpcode(), VectorTy, VectorType::get(Op0Ty, VectorTy), TTI::CastContextHint::None, CostKind, Op0); InstructionCost ExtCost1 = TTI.getCastInstrCost( Op1->getOpcode(), VectorTy, VectorType::get(Op1Ty, VectorTy), TTI::CastContextHint::None, CostKind, Op1); InstructionCost MulCost = TTI.getArithmeticInstrCost(Instruction::Mul, VectorTy, CostKind); InstructionCost RedCost = TTI.getMulAccReductionCost( IsUnsigned, RdxDesc.getRecurrenceType(), ExtType, CostKind); InstructionCost ExtraExtCost = 0; if (Op0Ty != LargestOpTy || Op1Ty != LargestOpTy) { Instruction *ExtraExtOp = (Op0Ty != LargestOpTy) ? Op0 : Op1; ExtraExtCost = TTI.getCastInstrCost( ExtraExtOp->getOpcode(), ExtType, VectorType::get(ExtraExtOp->getOperand(0)->getType(), VectorTy), TTI::CastContextHint::None, CostKind, ExtraExtOp); } if (RedCost.isValid() && (RedCost + ExtraExtCost) < (ExtCost0 + ExtCost1 + MulCost + BaseCost)) return I == RetI ? RedCost : 0; } else if (!match(I, m_ZExtOrSExt(m_Value()))) { // Matched reduce.add(mul()) InstructionCost MulCost = TTI.getArithmeticInstrCost(Instruction::Mul, VectorTy, CostKind); InstructionCost RedCost = TTI.getMulAccReductionCost( true, RdxDesc.getRecurrenceType(), VectorTy, CostKind); if (RedCost.isValid() && RedCost < MulCost + BaseCost) return I == RetI ? RedCost : 0; } } return I == RetI ? std::optional(BaseCost) : std::nullopt; } InstructionCost LoopVectorizationCostModel::getMemoryInstructionCost(Instruction *I, ElementCount VF) { // Calculate scalar cost only. Vectorization cost should be ready at this // moment. if (VF.isScalar()) { Type *ValTy = getLoadStoreType(I); const Align Alignment = getLoadStoreAlignment(I); unsigned AS = getLoadStoreAddressSpace(I); TTI::OperandValueInfo OpInfo = TTI::getOperandInfo(I->getOperand(0)); return TTI.getAddressComputationCost(ValTy) + TTI.getMemoryOpCost(I->getOpcode(), ValTy, Alignment, AS, TTI::TCK_RecipThroughput, OpInfo, I); } return getWideningCost(I, VF); } LoopVectorizationCostModel::VectorizationCostTy LoopVectorizationCostModel::getInstructionCost(Instruction *I, ElementCount VF) { // If we know that this instruction will remain uniform, check the cost of // the scalar version. if (isUniformAfterVectorization(I, VF)) VF = ElementCount::getFixed(1); if (VF.isVector() && isProfitableToScalarize(I, VF)) return VectorizationCostTy(InstsToScalarize[VF][I], false); // Forced scalars do not have any scalarization overhead. auto ForcedScalar = ForcedScalars.find(VF); if (VF.isVector() && ForcedScalar != ForcedScalars.end()) { auto InstSet = ForcedScalar->second; if (InstSet.count(I)) return VectorizationCostTy( (getInstructionCost(I, ElementCount::getFixed(1)).first * VF.getKnownMinValue()), false); } Type *VectorTy; InstructionCost C = getInstructionCost(I, VF, VectorTy); bool TypeNotScalarized = false; if (VF.isVector() && VectorTy->isVectorTy()) { if (unsigned NumParts = TTI.getNumberOfParts(VectorTy)) { if (VF.isScalable()) // is assumed to be profitable over iN because // scalable registers are a distinct register class from scalar ones. // If we ever find a target which wants to lower scalable vectors // back to scalars, we'll need to update this code to explicitly // ask TTI about the register class uses for each part. TypeNotScalarized = NumParts <= VF.getKnownMinValue(); else TypeNotScalarized = NumParts < VF.getKnownMinValue(); } else C = InstructionCost::getInvalid(); } return VectorizationCostTy(C, TypeNotScalarized); } InstructionCost LoopVectorizationCostModel::getScalarizationOverhead( Instruction *I, ElementCount VF, TTI::TargetCostKind CostKind) const { // There is no mechanism yet to create a scalable scalarization loop, // so this is currently Invalid. if (VF.isScalable()) return InstructionCost::getInvalid(); if (VF.isScalar()) return 0; InstructionCost Cost = 0; Type *RetTy = ToVectorTy(I->getType(), VF); if (!RetTy->isVoidTy() && (!isa(I) || !TTI.supportsEfficientVectorElementLoadStore())) Cost += TTI.getScalarizationOverhead( cast(RetTy), APInt::getAllOnes(VF.getKnownMinValue()), /*Insert*/ true, /*Extract*/ false, CostKind); // Some targets keep addresses scalar. if (isa(I) && !TTI.prefersVectorizedAddressing()) return Cost; // Some targets support efficient element stores. if (isa(I) && TTI.supportsEfficientVectorElementLoadStore()) return Cost; // Collect operands to consider. CallInst *CI = dyn_cast(I); Instruction::op_range Ops = CI ? CI->args() : I->operands(); // Skip operands that do not require extraction/scalarization and do not incur // any overhead. SmallVector Tys; for (auto *V : filterExtractingOperands(Ops, VF)) Tys.push_back(MaybeVectorizeType(V->getType(), VF)); return Cost + TTI.getOperandsScalarizationOverhead( filterExtractingOperands(Ops, VF), Tys, CostKind); } void LoopVectorizationCostModel::setCostBasedWideningDecision(ElementCount VF) { if (VF.isScalar()) return; NumPredStores = 0; for (BasicBlock *BB : TheLoop->blocks()) { // For each instruction in the old loop. for (Instruction &I : *BB) { Value *Ptr = getLoadStorePointerOperand(&I); if (!Ptr) continue; // TODO: We should generate better code and update the cost model for // predicated uniform stores. Today they are treated as any other // predicated store (see added test cases in // invariant-store-vectorization.ll). if (isa(&I) && isScalarWithPredication(&I, VF)) NumPredStores++; if (Legal->isUniformMemOp(I, VF)) { auto isLegalToScalarize = [&]() { if (!VF.isScalable()) // Scalarization of fixed length vectors "just works". return true; // We have dedicated lowering for unpredicated uniform loads and // stores. Note that even with tail folding we know that at least // one lane is active (i.e. generalized predication is not possible // here), and the logic below depends on this fact. if (!foldTailByMasking()) return true; // For scalable vectors, a uniform memop load is always // uniform-by-parts and we know how to scalarize that. if (isa(I)) return true; // A uniform store isn't neccessarily uniform-by-part // and we can't assume scalarization. auto &SI = cast(I); return TheLoop->isLoopInvariant(SI.getValueOperand()); }; const InstructionCost GatherScatterCost = isLegalGatherOrScatter(&I, VF) ? getGatherScatterCost(&I, VF) : InstructionCost::getInvalid(); // Load: Scalar load + broadcast // Store: Scalar store + isLoopInvariantStoreValue ? 0 : extract // FIXME: This cost is a significant under-estimate for tail folded // memory ops. const InstructionCost ScalarizationCost = isLegalToScalarize() ? getUniformMemOpCost(&I, VF) : InstructionCost::getInvalid(); // Choose better solution for the current VF, Note that Invalid // costs compare as maximumal large. If both are invalid, we get // scalable invalid which signals a failure and a vectorization abort. if (GatherScatterCost < ScalarizationCost) setWideningDecision(&I, VF, CM_GatherScatter, GatherScatterCost); else setWideningDecision(&I, VF, CM_Scalarize, ScalarizationCost); continue; } // We assume that widening is the best solution when possible. if (memoryInstructionCanBeWidened(&I, VF)) { InstructionCost Cost = getConsecutiveMemOpCost(&I, VF); int ConsecutiveStride = Legal->isConsecutivePtr( getLoadStoreType(&I), getLoadStorePointerOperand(&I)); assert((ConsecutiveStride == 1 || ConsecutiveStride == -1) && "Expected consecutive stride."); InstWidening Decision = ConsecutiveStride == 1 ? CM_Widen : CM_Widen_Reverse; setWideningDecision(&I, VF, Decision, Cost); continue; } // Choose between Interleaving, Gather/Scatter or Scalarization. InstructionCost InterleaveCost = InstructionCost::getInvalid(); unsigned NumAccesses = 1; if (isAccessInterleaved(&I)) { auto Group = getInterleavedAccessGroup(&I); assert(Group && "Fail to get an interleaved access group."); // Make one decision for the whole group. if (getWideningDecision(&I, VF) != CM_Unknown) continue; NumAccesses = Group->getNumMembers(); if (interleavedAccessCanBeWidened(&I, VF)) InterleaveCost = getInterleaveGroupCost(&I, VF); } InstructionCost GatherScatterCost = isLegalGatherOrScatter(&I, VF) ? getGatherScatterCost(&I, VF) * NumAccesses : InstructionCost::getInvalid(); InstructionCost ScalarizationCost = getMemInstScalarizationCost(&I, VF) * NumAccesses; // Choose better solution for the current VF, // write down this decision and use it during vectorization. InstructionCost Cost; InstWidening Decision; if (InterleaveCost <= GatherScatterCost && InterleaveCost < ScalarizationCost) { Decision = CM_Interleave; Cost = InterleaveCost; } else if (GatherScatterCost < ScalarizationCost) { Decision = CM_GatherScatter; Cost = GatherScatterCost; } else { Decision = CM_Scalarize; Cost = ScalarizationCost; } // If the instructions belongs to an interleave group, the whole group // receives the same decision. The whole group receives the cost, but // the cost will actually be assigned to one instruction. if (auto Group = getInterleavedAccessGroup(&I)) setWideningDecision(Group, VF, Decision, Cost); else setWideningDecision(&I, VF, Decision, Cost); } } // Make sure that any load of address and any other address computation // remains scalar unless there is gather/scatter support. This avoids // inevitable extracts into address registers, and also has the benefit of // activating LSR more, since that pass can't optimize vectorized // addresses. if (TTI.prefersVectorizedAddressing()) return; // Start with all scalar pointer uses. SmallPtrSet AddrDefs; for (BasicBlock *BB : TheLoop->blocks()) for (Instruction &I : *BB) { Instruction *PtrDef = dyn_cast_or_null(getLoadStorePointerOperand(&I)); if (PtrDef && TheLoop->contains(PtrDef) && getWideningDecision(&I, VF) != CM_GatherScatter) AddrDefs.insert(PtrDef); } // Add all instructions used to generate the addresses. SmallVector Worklist; append_range(Worklist, AddrDefs); while (!Worklist.empty()) { Instruction *I = Worklist.pop_back_val(); for (auto &Op : I->operands()) if (auto *InstOp = dyn_cast(Op)) if ((InstOp->getParent() == I->getParent()) && !isa(InstOp) && AddrDefs.insert(InstOp).second) Worklist.push_back(InstOp); } for (auto *I : AddrDefs) { if (isa(I)) { // Setting the desired widening decision should ideally be handled in // by cost functions, but since this involves the task of finding out // if the loaded register is involved in an address computation, it is // instead changed here when we know this is the case. InstWidening Decision = getWideningDecision(I, VF); if (Decision == CM_Widen || Decision == CM_Widen_Reverse) // Scalarize a widened load of address. setWideningDecision( I, VF, CM_Scalarize, (VF.getKnownMinValue() * getMemoryInstructionCost(I, ElementCount::getFixed(1)))); else if (auto Group = getInterleavedAccessGroup(I)) { // Scalarize an interleave group of address loads. for (unsigned I = 0; I < Group->getFactor(); ++I) { if (Instruction *Member = Group->getMember(I)) setWideningDecision( Member, VF, CM_Scalarize, (VF.getKnownMinValue() * getMemoryInstructionCost(Member, ElementCount::getFixed(1)))); } } } else // Make sure I gets scalarized and a cost estimate without // scalarization overhead. ForcedScalars[VF].insert(I); } } void LoopVectorizationCostModel::setVectorizedCallDecision(ElementCount VF) { assert(!VF.isScalar() && "Trying to set a vectorization decision for a scalar VF"); for (BasicBlock *BB : TheLoop->blocks()) { // For each instruction in the old loop. for (Instruction &I : *BB) { CallInst *CI = dyn_cast(&I); if (!CI) continue; InstructionCost ScalarCost = InstructionCost::getInvalid(); InstructionCost VectorCost = InstructionCost::getInvalid(); InstructionCost IntrinsicCost = InstructionCost::getInvalid(); TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput; Function *ScalarFunc = CI->getCalledFunction(); Type *ScalarRetTy = CI->getType(); SmallVector Tys, ScalarTys; bool MaskRequired = Legal->isMaskRequired(CI); for (auto &ArgOp : CI->args()) ScalarTys.push_back(ArgOp->getType()); // Compute corresponding vector type for return value and arguments. Type *RetTy = ToVectorTy(ScalarRetTy, VF); for (Type *ScalarTy : ScalarTys) Tys.push_back(ToVectorTy(ScalarTy, VF)); // An in-loop reduction using an fmuladd intrinsic is a special case; // we don't want the normal cost for that intrinsic. if (RecurrenceDescriptor::isFMulAddIntrinsic(CI)) if (auto RedCost = getReductionPatternCost(CI, VF, RetTy, CostKind)) { setCallWideningDecision(CI, VF, CM_IntrinsicCall, nullptr, getVectorIntrinsicIDForCall(CI, TLI), std::nullopt, *RedCost); continue; } // Estimate cost of scalarized vector call. The source operands are // assumed to be vectors, so we need to extract individual elements from // there, execute VF scalar calls, and then gather the result into the // vector return value. InstructionCost ScalarCallCost = TTI.getCallInstrCost(ScalarFunc, ScalarRetTy, ScalarTys, CostKind); // Compute costs of unpacking argument values for the scalar calls and // packing the return values to a vector. InstructionCost ScalarizationCost = getScalarizationOverhead(CI, VF, CostKind); ScalarCost = ScalarCallCost * VF.getKnownMinValue() + ScalarizationCost; // Find the cost of vectorizing the call, if we can find a suitable // vector variant of the function. bool UsesMask = false; VFInfo FuncInfo; Function *VecFunc = nullptr; // Search through any available variants for one we can use at this VF. for (VFInfo &Info : VFDatabase::getMappings(*CI)) { // Must match requested VF. if (Info.Shape.VF != VF) continue; // Must take a mask argument if one is required if (MaskRequired && !Info.isMasked()) continue; // Check that all parameter kinds are supported bool ParamsOk = true; for (VFParameter Param : Info.Shape.Parameters) { switch (Param.ParamKind) { case VFParamKind::Vector: break; case VFParamKind::OMP_Uniform: { Value *ScalarParam = CI->getArgOperand(Param.ParamPos); // Make sure the scalar parameter in the loop is invariant. if (!PSE.getSE()->isLoopInvariant(PSE.getSCEV(ScalarParam), TheLoop)) ParamsOk = false; break; } case VFParamKind::OMP_Linear: { Value *ScalarParam = CI->getArgOperand(Param.ParamPos); // Find the stride for the scalar parameter in this loop and see if // it matches the stride for the variant. // TODO: do we need to figure out the cost of an extract to get the // first lane? Or do we hope that it will be folded away? ScalarEvolution *SE = PSE.getSE(); const auto *SAR = dyn_cast(SE->getSCEV(ScalarParam)); if (!SAR || SAR->getLoop() != TheLoop) { ParamsOk = false; break; } const SCEVConstant *Step = dyn_cast(SAR->getStepRecurrence(*SE)); if (!Step || Step->getAPInt().getSExtValue() != Param.LinearStepOrPos) ParamsOk = false; break; } case VFParamKind::GlobalPredicate: UsesMask = true; break; default: ParamsOk = false; break; } } if (!ParamsOk) continue; // Found a suitable candidate, stop here. VecFunc = CI->getModule()->getFunction(Info.VectorName); FuncInfo = Info; break; } // Add in the cost of synthesizing a mask if one wasn't required. InstructionCost MaskCost = 0; if (VecFunc && UsesMask && !MaskRequired) MaskCost = TTI.getShuffleCost( TargetTransformInfo::SK_Broadcast, VectorType::get(IntegerType::getInt1Ty( VecFunc->getFunctionType()->getContext()), VF)); if (TLI && VecFunc && !CI->isNoBuiltin()) VectorCost = TTI.getCallInstrCost(nullptr, RetTy, Tys, CostKind) + MaskCost; // Find the cost of an intrinsic; some targets may have instructions that // perform the operation without needing an actual call. Intrinsic::ID IID = getVectorIntrinsicIDForCall(CI, TLI); if (IID != Intrinsic::not_intrinsic) IntrinsicCost = getVectorIntrinsicCost(CI, VF); InstructionCost Cost = ScalarCost; InstWidening Decision = CM_Scalarize; if (VectorCost <= Cost) { Cost = VectorCost; Decision = CM_VectorCall; } if (IntrinsicCost <= Cost) { Cost = IntrinsicCost; Decision = CM_IntrinsicCall; } setCallWideningDecision(CI, VF, Decision, VecFunc, IID, FuncInfo.getParamIndexForOptionalMask(), Cost); } } } InstructionCost LoopVectorizationCostModel::getInstructionCost(Instruction *I, ElementCount VF, Type *&VectorTy) { Type *RetTy = I->getType(); if (canTruncateToMinimalBitwidth(I, VF)) RetTy = IntegerType::get(RetTy->getContext(), MinBWs[I]); auto SE = PSE.getSE(); TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput; auto hasSingleCopyAfterVectorization = [this](Instruction *I, ElementCount VF) -> bool { if (VF.isScalar()) return true; auto Scalarized = InstsToScalarize.find(VF); assert(Scalarized != InstsToScalarize.end() && "VF not yet analyzed for scalarization profitability"); return !Scalarized->second.count(I) && llvm::all_of(I->users(), [&](User *U) { auto *UI = cast(U); return !Scalarized->second.count(UI); }); }; (void) hasSingleCopyAfterVectorization; if (isScalarAfterVectorization(I, VF)) { // With the exception of GEPs and PHIs, after scalarization there should // only be one copy of the instruction generated in the loop. This is // because the VF is either 1, or any instructions that need scalarizing // have already been dealt with by the time we get here. As a result, // it means we don't have to multiply the instruction cost by VF. assert(I->getOpcode() == Instruction::GetElementPtr || I->getOpcode() == Instruction::PHI || (I->getOpcode() == Instruction::BitCast && I->getType()->isPointerTy()) || hasSingleCopyAfterVectorization(I, VF)); VectorTy = RetTy; } else VectorTy = ToVectorTy(RetTy, VF); // TODO: We need to estimate the cost of intrinsic calls. switch (I->getOpcode()) { case Instruction::GetElementPtr: // We mark this instruction as zero-cost because the cost of GEPs in // vectorized code depends on whether the corresponding memory instruction // is scalarized or not. Therefore, we handle GEPs with the memory // instruction cost. return 0; case Instruction::Br: { // In cases of scalarized and predicated instructions, there will be VF // predicated blocks in the vectorized loop. Each branch around these // blocks requires also an extract of its vector compare i1 element. bool ScalarPredicatedBB = false; BranchInst *BI = cast(I); if (VF.isVector() && BI->isConditional() && (PredicatedBBsAfterVectorization[VF].count(BI->getSuccessor(0)) || PredicatedBBsAfterVectorization[VF].count(BI->getSuccessor(1)))) ScalarPredicatedBB = true; if (ScalarPredicatedBB) { // Not possible to scalarize scalable vector with predicated instructions. if (VF.isScalable()) return InstructionCost::getInvalid(); // Return cost for branches around scalarized and predicated blocks. auto *Vec_i1Ty = VectorType::get(IntegerType::getInt1Ty(RetTy->getContext()), VF); return ( TTI.getScalarizationOverhead( Vec_i1Ty, APInt::getAllOnes(VF.getFixedValue()), /*Insert*/ false, /*Extract*/ true, CostKind) + (TTI.getCFInstrCost(Instruction::Br, CostKind) * VF.getFixedValue())); } else if (I->getParent() == TheLoop->getLoopLatch() || VF.isScalar()) // The back-edge branch will remain, as will all scalar branches. return TTI.getCFInstrCost(Instruction::Br, CostKind); else // This branch will be eliminated by if-conversion. return 0; // Note: We currently assume zero cost for an unconditional branch inside // a predicated block since it will become a fall-through, although we // may decide in the future to call TTI for all branches. } case Instruction::PHI: { auto *Phi = cast(I); // First-order recurrences are replaced by vector shuffles inside the loop. if (VF.isVector() && Legal->isFixedOrderRecurrence(Phi)) { SmallVector Mask(VF.getKnownMinValue()); std::iota(Mask.begin(), Mask.end(), VF.getKnownMinValue() - 1); return TTI.getShuffleCost(TargetTransformInfo::SK_Splice, cast(VectorTy), Mask, CostKind, VF.getKnownMinValue() - 1); } // Phi nodes in non-header blocks (not inductions, reductions, etc.) are // converted into select instructions. We require N - 1 selects per phi // node, where N is the number of incoming values. if (VF.isVector() && Phi->getParent() != TheLoop->getHeader()) return (Phi->getNumIncomingValues() - 1) * TTI.getCmpSelInstrCost( Instruction::Select, ToVectorTy(Phi->getType(), VF), ToVectorTy(Type::getInt1Ty(Phi->getContext()), VF), CmpInst::BAD_ICMP_PREDICATE, CostKind); return TTI.getCFInstrCost(Instruction::PHI, CostKind); } case Instruction::UDiv: case Instruction::SDiv: case Instruction::URem: case Instruction::SRem: if (VF.isVector() && isPredicatedInst(I)) { const auto [ScalarCost, SafeDivisorCost] = getDivRemSpeculationCost(I, VF); return isDivRemScalarWithPredication(ScalarCost, SafeDivisorCost) ? ScalarCost : SafeDivisorCost; } // We've proven all lanes safe to speculate, fall through. [[fallthrough]]; case Instruction::Add: case Instruction::FAdd: case Instruction::Sub: case Instruction::FSub: case Instruction::Mul: case Instruction::FMul: case Instruction::FDiv: case Instruction::FRem: case Instruction::Shl: case Instruction::LShr: case Instruction::AShr: case Instruction::And: case Instruction::Or: case Instruction::Xor: { // If we're speculating on the stride being 1, the multiplication may // fold away. We can generalize this for all operations using the notion // of neutral elements. (TODO) if (I->getOpcode() == Instruction::Mul && (PSE.getSCEV(I->getOperand(0))->isOne() || PSE.getSCEV(I->getOperand(1))->isOne())) return 0; // Detect reduction patterns if (auto RedCost = getReductionPatternCost(I, VF, VectorTy, CostKind)) return *RedCost; // Certain instructions can be cheaper to vectorize if they have a constant // second vector operand. One example of this are shifts on x86. Value *Op2 = I->getOperand(1); auto Op2Info = TTI.getOperandInfo(Op2); if (Op2Info.Kind == TargetTransformInfo::OK_AnyValue && Legal->isInvariant(Op2)) Op2Info.Kind = TargetTransformInfo::OK_UniformValue; SmallVector Operands(I->operand_values()); auto InstrCost = TTI.getArithmeticInstrCost( I->getOpcode(), VectorTy, CostKind, {TargetTransformInfo::OK_AnyValue, TargetTransformInfo::OP_None}, Op2Info, Operands, I); // Some targets can replace frem with vector library calls. InstructionCost VecCallCost = InstructionCost::getInvalid(); if (I->getOpcode() == Instruction::FRem) { LibFunc Func; if (TLI->getLibFunc(I->getOpcode(), I->getType(), Func) && TLI->isFunctionVectorizable(TLI->getName(Func), VF)) { SmallVector OpTypes; for (auto &Op : I->operands()) OpTypes.push_back(Op->getType()); VecCallCost = TTI.getCallInstrCost(nullptr, VectorTy, OpTypes, CostKind); } } return std::min(InstrCost, VecCallCost); } case Instruction::FNeg: { return TTI.getArithmeticInstrCost( I->getOpcode(), VectorTy, CostKind, {TargetTransformInfo::OK_AnyValue, TargetTransformInfo::OP_None}, {TargetTransformInfo::OK_AnyValue, TargetTransformInfo::OP_None}, I->getOperand(0), I); } case Instruction::Select: { SelectInst *SI = cast(I); const SCEV *CondSCEV = SE->getSCEV(SI->getCondition()); bool ScalarCond = (SE->isLoopInvariant(CondSCEV, TheLoop)); const Value *Op0, *Op1; using namespace llvm::PatternMatch; if (!ScalarCond && (match(I, m_LogicalAnd(m_Value(Op0), m_Value(Op1))) || match(I, m_LogicalOr(m_Value(Op0), m_Value(Op1))))) { // select x, y, false --> x & y // select x, true, y --> x | y const auto [Op1VK, Op1VP] = TTI::getOperandInfo(Op0); const auto [Op2VK, Op2VP] = TTI::getOperandInfo(Op1); assert(Op0->getType()->getScalarSizeInBits() == 1 && Op1->getType()->getScalarSizeInBits() == 1); SmallVector Operands{Op0, Op1}; return TTI.getArithmeticInstrCost( match(I, m_LogicalOr()) ? Instruction::Or : Instruction::And, VectorTy, CostKind, {Op1VK, Op1VP}, {Op2VK, Op2VP}, Operands, I); } Type *CondTy = SI->getCondition()->getType(); if (!ScalarCond) CondTy = VectorType::get(CondTy, VF); CmpInst::Predicate Pred = CmpInst::BAD_ICMP_PREDICATE; if (auto *Cmp = dyn_cast(SI->getCondition())) Pred = Cmp->getPredicate(); return TTI.getCmpSelInstrCost(I->getOpcode(), VectorTy, CondTy, Pred, CostKind, I); } case Instruction::ICmp: case Instruction::FCmp: { Type *ValTy = I->getOperand(0)->getType(); Instruction *Op0AsInstruction = dyn_cast(I->getOperand(0)); if (canTruncateToMinimalBitwidth(Op0AsInstruction, VF)) ValTy = IntegerType::get(ValTy->getContext(), MinBWs[Op0AsInstruction]); VectorTy = ToVectorTy(ValTy, VF); return TTI.getCmpSelInstrCost(I->getOpcode(), VectorTy, nullptr, cast(I)->getPredicate(), CostKind, I); } case Instruction::Store: case Instruction::Load: { ElementCount Width = VF; if (Width.isVector()) { InstWidening Decision = getWideningDecision(I, Width); assert(Decision != CM_Unknown && "CM decision should be taken at this point"); if (getWideningCost(I, VF) == InstructionCost::getInvalid()) return InstructionCost::getInvalid(); if (Decision == CM_Scalarize) Width = ElementCount::getFixed(1); } VectorTy = ToVectorTy(getLoadStoreType(I), Width); return getMemoryInstructionCost(I, VF); } case Instruction::BitCast: if (I->getType()->isPointerTy()) return 0; [[fallthrough]]; case Instruction::ZExt: case Instruction::SExt: case Instruction::FPToUI: case Instruction::FPToSI: case Instruction::FPExt: case Instruction::PtrToInt: case Instruction::IntToPtr: case Instruction::SIToFP: case Instruction::UIToFP: case Instruction::Trunc: case Instruction::FPTrunc: { // Computes the CastContextHint from a Load/Store instruction. auto ComputeCCH = [&](Instruction *I) -> TTI::CastContextHint { assert((isa(I) || isa(I)) && "Expected a load or a store!"); if (VF.isScalar() || !TheLoop->contains(I)) return TTI::CastContextHint::Normal; switch (getWideningDecision(I, VF)) { case LoopVectorizationCostModel::CM_GatherScatter: return TTI::CastContextHint::GatherScatter; case LoopVectorizationCostModel::CM_Interleave: return TTI::CastContextHint::Interleave; case LoopVectorizationCostModel::CM_Scalarize: case LoopVectorizationCostModel::CM_Widen: return Legal->isMaskRequired(I) ? TTI::CastContextHint::Masked : TTI::CastContextHint::Normal; case LoopVectorizationCostModel::CM_Widen_Reverse: return TTI::CastContextHint::Reversed; case LoopVectorizationCostModel::CM_Unknown: llvm_unreachable("Instr did not go through cost modelling?"); case LoopVectorizationCostModel::CM_VectorCall: case LoopVectorizationCostModel::CM_IntrinsicCall: llvm_unreachable_internal("Instr has invalid widening decision"); } llvm_unreachable("Unhandled case!"); }; unsigned Opcode = I->getOpcode(); TTI::CastContextHint CCH = TTI::CastContextHint::None; // For Trunc, the context is the only user, which must be a StoreInst. if (Opcode == Instruction::Trunc || Opcode == Instruction::FPTrunc) { if (I->hasOneUse()) if (StoreInst *Store = dyn_cast(*I->user_begin())) CCH = ComputeCCH(Store); } // For Z/Sext, the context is the operand, which must be a LoadInst. else if (Opcode == Instruction::ZExt || Opcode == Instruction::SExt || Opcode == Instruction::FPExt) { if (LoadInst *Load = dyn_cast(I->getOperand(0))) CCH = ComputeCCH(Load); } // We optimize the truncation of induction variables having constant // integer steps. The cost of these truncations is the same as the scalar // operation. if (isOptimizableIVTruncate(I, VF)) { auto *Trunc = cast(I); return TTI.getCastInstrCost(Instruction::Trunc, Trunc->getDestTy(), Trunc->getSrcTy(), CCH, CostKind, Trunc); } // Detect reduction patterns if (auto RedCost = getReductionPatternCost(I, VF, VectorTy, CostKind)) return *RedCost; Type *SrcScalarTy = I->getOperand(0)->getType(); Type *SrcVecTy = VectorTy->isVectorTy() ? ToVectorTy(SrcScalarTy, VF) : SrcScalarTy; if (canTruncateToMinimalBitwidth(I, VF)) { // This cast is going to be shrunk. This may remove the cast or it might // turn it into slightly different cast. For example, if MinBW == 16, // "zext i8 %1 to i32" becomes "zext i8 %1 to i16". // // Calculate the modified src and dest types. Type *MinVecTy = VectorTy; if (Opcode == Instruction::Trunc) { SrcVecTy = smallestIntegerVectorType(SrcVecTy, MinVecTy); VectorTy = largestIntegerVectorType(ToVectorTy(I->getType(), VF), MinVecTy); } else if (Opcode == Instruction::ZExt || Opcode == Instruction::SExt) { // Leave SrcVecTy unchanged - we only shrink the destination element // type. VectorTy = smallestIntegerVectorType(ToVectorTy(I->getType(), VF), MinVecTy); } } return TTI.getCastInstrCost(Opcode, VectorTy, SrcVecTy, CCH, CostKind, I); } case Instruction::Call: return getVectorCallCost(cast(I), VF); case Instruction::ExtractValue: return TTI.getInstructionCost(I, TTI::TCK_RecipThroughput); case Instruction::Alloca: // We cannot easily widen alloca to a scalable alloca, as // the result would need to be a vector of pointers. if (VF.isScalable()) return InstructionCost::getInvalid(); [[fallthrough]]; default: // This opcode is unknown. Assume that it is the same as 'mul'. return TTI.getArithmeticInstrCost(Instruction::Mul, VectorTy, CostKind); } // end of switch. } void LoopVectorizationCostModel::collectValuesToIgnore() { // Ignore ephemeral values. CodeMetrics::collectEphemeralValues(TheLoop, AC, ValuesToIgnore); // Find all stores to invariant variables. Since they are going to sink // outside the loop we do not need calculate cost for them. for (BasicBlock *BB : TheLoop->blocks()) for (Instruction &I : *BB) { StoreInst *SI; if ((SI = dyn_cast(&I)) && Legal->isInvariantAddressOfReduction(SI->getPointerOperand())) ValuesToIgnore.insert(&I); } // Ignore type-promoting instructions we identified during reduction // detection. for (const auto &Reduction : Legal->getReductionVars()) { const RecurrenceDescriptor &RedDes = Reduction.second; const SmallPtrSetImpl &Casts = RedDes.getCastInsts(); VecValuesToIgnore.insert(Casts.begin(), Casts.end()); } // Ignore type-casting instructions we identified during induction // detection. for (const auto &Induction : Legal->getInductionVars()) { const InductionDescriptor &IndDes = Induction.second; const SmallVectorImpl &Casts = IndDes.getCastInsts(); VecValuesToIgnore.insert(Casts.begin(), Casts.end()); } } void LoopVectorizationCostModel::collectInLoopReductions() { for (const auto &Reduction : Legal->getReductionVars()) { PHINode *Phi = Reduction.first; const RecurrenceDescriptor &RdxDesc = Reduction.second; // We don't collect reductions that are type promoted (yet). if (RdxDesc.getRecurrenceType() != Phi->getType()) continue; // If the target would prefer this reduction to happen "in-loop", then we // want to record it as such. unsigned Opcode = RdxDesc.getOpcode(); if (!PreferInLoopReductions && !useOrderedReductions(RdxDesc) && !TTI.preferInLoopReduction(Opcode, Phi->getType(), TargetTransformInfo::ReductionFlags())) continue; // Check that we can correctly put the reductions into the loop, by // finding the chain of operations that leads from the phi to the loop // exit value. SmallVector ReductionOperations = RdxDesc.getReductionOpChain(Phi, TheLoop); bool InLoop = !ReductionOperations.empty(); if (InLoop) { InLoopReductions.insert(Phi); // Add the elements to InLoopReductionImmediateChains for cost modelling. Instruction *LastChain = Phi; for (auto *I : ReductionOperations) { InLoopReductionImmediateChains[I] = LastChain; LastChain = I; } } LLVM_DEBUG(dbgs() << "LV: Using " << (InLoop ? "inloop" : "out of loop") << " reduction for phi: " << *Phi << "\n"); } } VPValue *VPBuilder::createICmp(CmpInst::Predicate Pred, VPValue *A, VPValue *B, DebugLoc DL, const Twine &Name) { assert(Pred >= CmpInst::FIRST_ICMP_PREDICATE && Pred <= CmpInst::LAST_ICMP_PREDICATE && "invalid predicate"); return tryInsertInstruction( new VPInstruction(Instruction::ICmp, Pred, A, B, DL, Name)); } // This function will select a scalable VF if the target supports scalable // vectors and a fixed one otherwise. // TODO: we could return a pair of values that specify the max VF and // min VF, to be used in `buildVPlans(MinVF, MaxVF)` instead of // `buildVPlans(VF, VF)`. We cannot do it because VPLAN at the moment // doesn't have a cost model that can choose which plan to execute if // more than one is generated. static ElementCount determineVPlanVF(const TargetTransformInfo &TTI, LoopVectorizationCostModel &CM) { unsigned WidestType; std::tie(std::ignore, WidestType) = CM.getSmallestAndWidestTypes(); TargetTransformInfo::RegisterKind RegKind = TTI.enableScalableVectorization() ? TargetTransformInfo::RGK_ScalableVector : TargetTransformInfo::RGK_FixedWidthVector; TypeSize RegSize = TTI.getRegisterBitWidth(RegKind); unsigned N = RegSize.getKnownMinValue() / WidestType; return ElementCount::get(N, RegSize.isScalable()); } VectorizationFactor LoopVectorizationPlanner::planInVPlanNativePath(ElementCount UserVF) { ElementCount VF = UserVF; // Outer loop handling: They may require CFG and instruction level // transformations before even evaluating whether vectorization is profitable. // Since we cannot modify the incoming IR, we need to build VPlan upfront in // the vectorization pipeline. if (!OrigLoop->isInnermost()) { // If the user doesn't provide a vectorization factor, determine a // reasonable one. if (UserVF.isZero()) { VF = determineVPlanVF(TTI, CM); LLVM_DEBUG(dbgs() << "LV: VPlan computed VF " << VF << ".\n"); // Make sure we have a VF > 1 for stress testing. if (VPlanBuildStressTest && (VF.isScalar() || VF.isZero())) { LLVM_DEBUG(dbgs() << "LV: VPlan stress testing: " << "overriding computed VF.\n"); VF = ElementCount::getFixed(4); } } else if (UserVF.isScalable() && !TTI.supportsScalableVectors() && !ForceTargetSupportsScalableVectors) { LLVM_DEBUG(dbgs() << "LV: Not vectorizing. Scalable VF requested, but " << "not supported by the target.\n"); reportVectorizationFailure( "Scalable vectorization requested but not supported by the target", "the scalable user-specified vectorization width for outer-loop " "vectorization cannot be used because the target does not support " "scalable vectors.", "ScalableVFUnfeasible", ORE, OrigLoop); return VectorizationFactor::Disabled(); } assert(EnableVPlanNativePath && "VPlan-native path is not enabled."); assert(isPowerOf2_32(VF.getKnownMinValue()) && "VF needs to be a power of two"); LLVM_DEBUG(dbgs() << "LV: Using " << (!UserVF.isZero() ? "user " : "") << "VF " << VF << " to build VPlans.\n"); buildVPlans(VF, VF); // For VPlan build stress testing, we bail out after VPlan construction. if (VPlanBuildStressTest) return VectorizationFactor::Disabled(); return {VF, 0 /*Cost*/, 0 /* ScalarCost */}; } LLVM_DEBUG( dbgs() << "LV: Not vectorizing. Inner loops aren't supported in the " "VPlan-native path.\n"); return VectorizationFactor::Disabled(); } std::optional LoopVectorizationPlanner::plan(ElementCount UserVF, unsigned UserIC) { assert(OrigLoop->isInnermost() && "Inner loop expected."); CM.collectValuesToIgnore(); CM.collectElementTypesForWidening(); FixedScalableVFPair MaxFactors = CM.computeMaxVF(UserVF, UserIC); if (!MaxFactors) // Cases that should not to be vectorized nor interleaved. return std::nullopt; // Invalidate interleave groups if all blocks of loop will be predicated. if (CM.blockNeedsPredicationForAnyReason(OrigLoop->getHeader()) && !useMaskedInterleavedAccesses(TTI)) { LLVM_DEBUG( dbgs() << "LV: Invalidate all interleaved groups due to fold-tail by masking " "which requires masked-interleaved support.\n"); if (CM.InterleaveInfo.invalidateGroups()) // Invalidating interleave groups also requires invalidating all decisions // based on them, which includes widening decisions and uniform and scalar // values. CM.invalidateCostModelingDecisions(); } ElementCount MaxUserVF = UserVF.isScalable() ? MaxFactors.ScalableVF : MaxFactors.FixedVF; bool UserVFIsLegal = ElementCount::isKnownLE(UserVF, MaxUserVF); if (!UserVF.isZero() && UserVFIsLegal) { assert(isPowerOf2_32(UserVF.getKnownMinValue()) && "VF needs to be a power of two"); // Collect the instructions (and their associated costs) that will be more // profitable to scalarize. CM.collectInLoopReductions(); if (CM.selectUserVectorizationFactor(UserVF)) { LLVM_DEBUG(dbgs() << "LV: Using user VF " << UserVF << ".\n"); buildVPlansWithVPRecipes(UserVF, UserVF); if (!hasPlanWithVF(UserVF)) { LLVM_DEBUG(dbgs() << "LV: No VPlan could be built for " << UserVF << ".\n"); return std::nullopt; } LLVM_DEBUG(printPlans(dbgs())); return {{UserVF, 0, 0}}; } else reportVectorizationInfo("UserVF ignored because of invalid costs.", "InvalidCost", ORE, OrigLoop); } // Populate the set of Vectorization Factor Candidates. ElementCountSet VFCandidates; for (auto VF = ElementCount::getFixed(1); ElementCount::isKnownLE(VF, MaxFactors.FixedVF); VF *= 2) VFCandidates.insert(VF); for (auto VF = ElementCount::getScalable(1); ElementCount::isKnownLE(VF, MaxFactors.ScalableVF); VF *= 2) VFCandidates.insert(VF); CM.collectInLoopReductions(); for (const auto &VF : VFCandidates) { // Collect Uniform and Scalar instructions after vectorization with VF. CM.collectUniformsAndScalars(VF); // Collect the instructions (and their associated costs) that will be more // profitable to scalarize. if (VF.isVector()) CM.collectInstsToScalarize(VF); } buildVPlansWithVPRecipes(ElementCount::getFixed(1), MaxFactors.FixedVF); buildVPlansWithVPRecipes(ElementCount::getScalable(1), MaxFactors.ScalableVF); LLVM_DEBUG(printPlans(dbgs())); if (!MaxFactors.hasVector()) return VectorizationFactor::Disabled(); // Select the optimal vectorization factor. VectorizationFactor VF = selectVectorizationFactor(VFCandidates); assert((VF.Width.isScalar() || VF.ScalarCost > 0) && "when vectorizing, the scalar cost must be non-zero."); if (!hasPlanWithVF(VF.Width)) { LLVM_DEBUG(dbgs() << "LV: No VPlan could be built for " << VF.Width << ".\n"); return std::nullopt; } return VF; } VPlan &LoopVectorizationPlanner::getBestPlanFor(ElementCount VF) const { assert(count_if(VPlans, [VF](const VPlanPtr &Plan) { return Plan->hasVF(VF); }) == 1 && "Best VF has not a single VPlan."); for (const VPlanPtr &Plan : VPlans) { if (Plan->hasVF(VF)) return *Plan.get(); } llvm_unreachable("No plan found!"); } static void AddRuntimeUnrollDisableMetaData(Loop *L) { SmallVector MDs; // Reserve first location for self reference to the LoopID metadata node. MDs.push_back(nullptr); bool IsUnrollMetadata = false; MDNode *LoopID = L->getLoopID(); if (LoopID) { // First find existing loop unrolling disable metadata. for (unsigned i = 1, ie = LoopID->getNumOperands(); i < ie; ++i) { auto *MD = dyn_cast(LoopID->getOperand(i)); if (MD) { const auto *S = dyn_cast(MD->getOperand(0)); IsUnrollMetadata = S && S->getString().starts_with("llvm.loop.unroll.disable"); } MDs.push_back(LoopID->getOperand(i)); } } if (!IsUnrollMetadata) { // Add runtime unroll disable metadata. LLVMContext &Context = L->getHeader()->getContext(); SmallVector DisableOperands; DisableOperands.push_back( MDString::get(Context, "llvm.loop.unroll.runtime.disable")); MDNode *DisableNode = MDNode::get(Context, DisableOperands); MDs.push_back(DisableNode); MDNode *NewLoopID = MDNode::get(Context, MDs); // Set operand 0 to refer to the loop id itself. NewLoopID->replaceOperandWith(0, NewLoopID); L->setLoopID(NewLoopID); } } // Check if \p RedResult is a ComputeReductionResult instruction, and if it is // create a merge phi node for it and add it to \p ReductionResumeValues. static void createAndCollectMergePhiForReduction( VPInstruction *RedResult, DenseMap &ReductionResumeValues, VPTransformState &State, Loop *OrigLoop, BasicBlock *LoopMiddleBlock) { if (!RedResult || RedResult->getOpcode() != VPInstruction::ComputeReductionResult) return; auto *PhiR = cast(RedResult->getOperand(0)); const RecurrenceDescriptor &RdxDesc = PhiR->getRecurrenceDescriptor(); TrackingVH ReductionStartValue = RdxDesc.getRecurrenceStartValue(); Value *FinalValue = State.get(RedResult, VPIteration(State.UF - 1, VPLane::getFirstLane())); auto *ResumePhi = dyn_cast(PhiR->getStartValue()->getUnderlyingValue()); // TODO: bc.merge.rdx should not be created here, instead it should be // modeled in VPlan. BasicBlock *LoopScalarPreHeader = OrigLoop->getLoopPreheader(); // Create a phi node that merges control-flow from the backedge-taken check // block and the middle block. auto *BCBlockPhi = PHINode::Create(FinalValue->getType(), 2, "bc.merge.rdx", LoopScalarPreHeader->getTerminator()); // If we are fixing reductions in the epilogue loop then we should already // have created a bc.merge.rdx Phi after the main vector body. Ensure that // we carry over the incoming values correctly. for (auto *Incoming : predecessors(LoopScalarPreHeader)) { if (Incoming == LoopMiddleBlock) BCBlockPhi->addIncoming(FinalValue, Incoming); else if (ResumePhi && is_contained(ResumePhi->blocks(), Incoming)) BCBlockPhi->addIncoming(ResumePhi->getIncomingValueForBlock(Incoming), Incoming); else BCBlockPhi->addIncoming(ReductionStartValue, Incoming); } auto *OrigPhi = cast(PhiR->getUnderlyingValue()); // TODO: This fixup should instead be modeled in VPlan. // Fix the scalar loop reduction variable with the incoming reduction sum // from the vector body and from the backedge value. int IncomingEdgeBlockIdx = OrigPhi->getBasicBlockIndex(OrigLoop->getLoopLatch()); assert(IncomingEdgeBlockIdx >= 0 && "Invalid block index"); // Pick the other block. int SelfEdgeBlockIdx = (IncomingEdgeBlockIdx ? 0 : 1); OrigPhi->setIncomingValue(SelfEdgeBlockIdx, BCBlockPhi); Instruction *LoopExitInst = RdxDesc.getLoopExitInstr(); OrigPhi->setIncomingValue(IncomingEdgeBlockIdx, LoopExitInst); ReductionResumeValues[&RdxDesc] = BCBlockPhi; } std::pair, DenseMap> LoopVectorizationPlanner::executePlan( ElementCount BestVF, unsigned BestUF, VPlan &BestVPlan, InnerLoopVectorizer &ILV, DominatorTree *DT, bool IsEpilogueVectorization, const DenseMap *ExpandedSCEVs) { assert(BestVPlan.hasVF(BestVF) && "Trying to execute plan with unsupported VF"); assert(BestVPlan.hasUF(BestUF) && "Trying to execute plan with unsupported UF"); assert( (IsEpilogueVectorization || !ExpandedSCEVs) && "expanded SCEVs to reuse can only be used during epilogue vectorization"); LLVM_DEBUG(dbgs() << "Executing best plan with VF=" << BestVF << ", UF=" << BestUF << '\n'); if (!IsEpilogueVectorization) VPlanTransforms::optimizeForVFAndUF(BestVPlan, BestVF, BestUF, PSE); // Perform the actual loop transformation. VPTransformState State(BestVF, BestUF, LI, DT, ILV.Builder, &ILV, &BestVPlan, OrigLoop->getHeader()->getContext()); // 0. Generate SCEV-dependent code into the preheader, including TripCount, // before making any changes to the CFG. if (!BestVPlan.getPreheader()->empty()) { State.CFG.PrevBB = OrigLoop->getLoopPreheader(); State.Builder.SetInsertPoint(OrigLoop->getLoopPreheader()->getTerminator()); BestVPlan.getPreheader()->execute(&State); } if (!ILV.getTripCount()) ILV.setTripCount(State.get(BestVPlan.getTripCount(), {0, 0})); else assert(IsEpilogueVectorization && "should only re-use the existing trip " "count during epilogue vectorization"); // 1. Set up the skeleton for vectorization, including vector pre-header and // middle block. The vector loop is created during VPlan execution. Value *CanonicalIVStartValue; std::tie(State.CFG.PrevBB, CanonicalIVStartValue) = ILV.createVectorizedLoopSkeleton(ExpandedSCEVs ? *ExpandedSCEVs : State.ExpandedSCEVs); // Only use noalias metadata when using memory checks guaranteeing no overlap // across all iterations. const LoopAccessInfo *LAI = ILV.Legal->getLAI(); std::unique_ptr LVer = nullptr; if (LAI && !LAI->getRuntimePointerChecking()->getChecks().empty() && !LAI->getRuntimePointerChecking()->getDiffChecks()) { // We currently don't use LoopVersioning for the actual loop cloning but we // still use it to add the noalias metadata. // TODO: Find a better way to re-use LoopVersioning functionality to add // metadata. LVer = std::make_unique( *LAI, LAI->getRuntimePointerChecking()->getChecks(), OrigLoop, LI, DT, PSE.getSE()); State.LVer = &*LVer; State.LVer->prepareNoAliasMetadata(); } ILV.collectPoisonGeneratingRecipes(State); ILV.printDebugTracesAtStart(); //===------------------------------------------------===// // // Notice: any optimization or new instruction that go // into the code below should also be implemented in // the cost-model. // //===------------------------------------------------===// // 2. Copy and widen instructions from the old loop into the new loop. BestVPlan.prepareToExecute(ILV.getTripCount(), ILV.getOrCreateVectorTripCount(nullptr), CanonicalIVStartValue, State); BestVPlan.execute(&State); // 2.5 Collect reduction resume values. DenseMap ReductionResumeValues; auto *ExitVPBB = cast(BestVPlan.getVectorLoopRegion()->getSingleSuccessor()); for (VPRecipeBase &R : *ExitVPBB) { createAndCollectMergePhiForReduction(dyn_cast(&R), ReductionResumeValues, State, OrigLoop, State.CFG.VPBB2IRBB[ExitVPBB]); } // 2.6. Maintain Loop Hints // Keep all loop hints from the original loop on the vector loop (we'll // replace the vectorizer-specific hints below). MDNode *OrigLoopID = OrigLoop->getLoopID(); std::optional VectorizedLoopID = makeFollowupLoopID(OrigLoopID, {LLVMLoopVectorizeFollowupAll, LLVMLoopVectorizeFollowupVectorized}); VPBasicBlock *HeaderVPBB = BestVPlan.getVectorLoopRegion()->getEntryBasicBlock(); Loop *L = LI->getLoopFor(State.CFG.VPBB2IRBB[HeaderVPBB]); if (VectorizedLoopID) L->setLoopID(*VectorizedLoopID); else { // Keep all loop hints from the original loop on the vector loop (we'll // replace the vectorizer-specific hints below). if (MDNode *LID = OrigLoop->getLoopID()) L->setLoopID(LID); LoopVectorizeHints Hints(L, true, *ORE); Hints.setAlreadyVectorized(); } TargetTransformInfo::UnrollingPreferences UP; TTI.getUnrollingPreferences(L, *PSE.getSE(), UP, ORE); if (!UP.UnrollVectorizedLoop || CanonicalIVStartValue) AddRuntimeUnrollDisableMetaData(L); // 3. Fix the vectorized code: take care of header phi's, live-outs, // predication, updating analyses. ILV.fixVectorizedLoop(State, BestVPlan); ILV.printDebugTracesAtEnd(); return {State.ExpandedSCEVs, ReductionResumeValues}; } #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP) void LoopVectorizationPlanner::printPlans(raw_ostream &O) { for (const auto &Plan : VPlans) if (PrintVPlansInDotFormat) Plan->printDOT(O); else Plan->print(O); } #endif //===--------------------------------------------------------------------===// // EpilogueVectorizerMainLoop //===--------------------------------------------------------------------===// /// This function is partially responsible for generating the control flow /// depicted in https://llvm.org/docs/Vectorizers.html#epilogue-vectorization. std::pair EpilogueVectorizerMainLoop::createEpilogueVectorizedLoopSkeleton( const SCEV2ValueTy &ExpandedSCEVs) { createVectorLoopSkeleton(""); // Generate the code to check the minimum iteration count of the vector // epilogue (see below). EPI.EpilogueIterationCountCheck = emitIterationCountCheck(LoopScalarPreHeader, true); EPI.EpilogueIterationCountCheck->setName("iter.check"); // Generate the code to check any assumptions that we've made for SCEV // expressions. EPI.SCEVSafetyCheck = emitSCEVChecks(LoopScalarPreHeader); // Generate the code that checks at runtime if arrays overlap. We put the // checks into a separate block to make the more common case of few elements // faster. EPI.MemSafetyCheck = emitMemRuntimeChecks(LoopScalarPreHeader); // Generate the iteration count check for the main loop, *after* the check // for the epilogue loop, so that the path-length is shorter for the case // that goes directly through the vector epilogue. The longer-path length for // the main loop is compensated for, by the gain from vectorizing the larger // trip count. Note: the branch will get updated later on when we vectorize // the epilogue. EPI.MainLoopIterationCountCheck = emitIterationCountCheck(LoopScalarPreHeader, false); // Generate the induction variable. EPI.VectorTripCount = getOrCreateVectorTripCount(LoopVectorPreHeader); // Skip induction resume value creation here because they will be created in // the second pass for the scalar loop. The induction resume values for the // inductions in the epilogue loop are created before executing the plan for // the epilogue loop. return {completeLoopSkeleton(), nullptr}; } void EpilogueVectorizerMainLoop::printDebugTracesAtStart() { LLVM_DEBUG({ dbgs() << "Create Skeleton for epilogue vectorized loop (first pass)\n" << "Main Loop VF:" << EPI.MainLoopVF << ", Main Loop UF:" << EPI.MainLoopUF << ", Epilogue Loop VF:" << EPI.EpilogueVF << ", Epilogue Loop UF:" << EPI.EpilogueUF << "\n"; }); } void EpilogueVectorizerMainLoop::printDebugTracesAtEnd() { DEBUG_WITH_TYPE(VerboseDebug, { dbgs() << "intermediate fn:\n" << *OrigLoop->getHeader()->getParent() << "\n"; }); } BasicBlock * EpilogueVectorizerMainLoop::emitIterationCountCheck(BasicBlock *Bypass, bool ForEpilogue) { assert(Bypass && "Expected valid bypass basic block."); ElementCount VFactor = ForEpilogue ? EPI.EpilogueVF : VF; unsigned UFactor = ForEpilogue ? EPI.EpilogueUF : UF; Value *Count = getTripCount(); // Reuse existing vector loop preheader for TC checks. // Note that new preheader block is generated for vector loop. BasicBlock *const TCCheckBlock = LoopVectorPreHeader; IRBuilder<> Builder(TCCheckBlock->getTerminator()); // Generate code to check if the loop's trip count is less than VF * UF of the // main vector loop. auto P = Cost->requiresScalarEpilogue(ForEpilogue ? EPI.EpilogueVF.isVector() : VF.isVector()) ? ICmpInst::ICMP_ULE : ICmpInst::ICMP_ULT; Value *CheckMinIters = Builder.CreateICmp( P, Count, createStepForVF(Builder, Count->getType(), VFactor, UFactor), "min.iters.check"); if (!ForEpilogue) TCCheckBlock->setName("vector.main.loop.iter.check"); // Create new preheader for vector loop. LoopVectorPreHeader = SplitBlock(TCCheckBlock, TCCheckBlock->getTerminator(), DT, LI, nullptr, "vector.ph"); if (ForEpilogue) { assert(DT->properlyDominates(DT->getNode(TCCheckBlock), DT->getNode(Bypass)->getIDom()) && "TC check is expected to dominate Bypass"); // Update dominator for Bypass & LoopExit. DT->changeImmediateDominator(Bypass, TCCheckBlock); if (!Cost->requiresScalarEpilogue(EPI.EpilogueVF.isVector())) // For loops with multiple exits, there's no edge from the middle block // to exit blocks (as the epilogue must run) and thus no need to update // the immediate dominator of the exit blocks. DT->changeImmediateDominator(LoopExitBlock, TCCheckBlock); LoopBypassBlocks.push_back(TCCheckBlock); // Save the trip count so we don't have to regenerate it in the // vec.epilog.iter.check. This is safe to do because the trip count // generated here dominates the vector epilog iter check. EPI.TripCount = Count; } BranchInst &BI = *BranchInst::Create(Bypass, LoopVectorPreHeader, CheckMinIters); if (hasBranchWeightMD(*OrigLoop->getLoopLatch()->getTerminator())) setBranchWeights(BI, MinItersBypassWeights); ReplaceInstWithInst(TCCheckBlock->getTerminator(), &BI); return TCCheckBlock; } //===--------------------------------------------------------------------===// // EpilogueVectorizerEpilogueLoop //===--------------------------------------------------------------------===// /// This function is partially responsible for generating the control flow /// depicted in https://llvm.org/docs/Vectorizers.html#epilogue-vectorization. std::pair EpilogueVectorizerEpilogueLoop::createEpilogueVectorizedLoopSkeleton( const SCEV2ValueTy &ExpandedSCEVs) { createVectorLoopSkeleton("vec.epilog."); // Now, compare the remaining count and if there aren't enough iterations to // execute the vectorized epilogue skip to the scalar part. BasicBlock *VecEpilogueIterationCountCheck = LoopVectorPreHeader; VecEpilogueIterationCountCheck->setName("vec.epilog.iter.check"); LoopVectorPreHeader = SplitBlock(LoopVectorPreHeader, LoopVectorPreHeader->getTerminator(), DT, LI, nullptr, "vec.epilog.ph"); emitMinimumVectorEpilogueIterCountCheck(LoopScalarPreHeader, VecEpilogueIterationCountCheck); // Adjust the control flow taking the state info from the main loop // vectorization into account. assert(EPI.MainLoopIterationCountCheck && EPI.EpilogueIterationCountCheck && "expected this to be saved from the previous pass."); EPI.MainLoopIterationCountCheck->getTerminator()->replaceUsesOfWith( VecEpilogueIterationCountCheck, LoopVectorPreHeader); DT->changeImmediateDominator(LoopVectorPreHeader, EPI.MainLoopIterationCountCheck); EPI.EpilogueIterationCountCheck->getTerminator()->replaceUsesOfWith( VecEpilogueIterationCountCheck, LoopScalarPreHeader); if (EPI.SCEVSafetyCheck) EPI.SCEVSafetyCheck->getTerminator()->replaceUsesOfWith( VecEpilogueIterationCountCheck, LoopScalarPreHeader); if (EPI.MemSafetyCheck) EPI.MemSafetyCheck->getTerminator()->replaceUsesOfWith( VecEpilogueIterationCountCheck, LoopScalarPreHeader); DT->changeImmediateDominator( VecEpilogueIterationCountCheck, VecEpilogueIterationCountCheck->getSinglePredecessor()); DT->changeImmediateDominator(LoopScalarPreHeader, EPI.EpilogueIterationCountCheck); if (!Cost->requiresScalarEpilogue(EPI.EpilogueVF.isVector())) // If there is an epilogue which must run, there's no edge from the // middle block to exit blocks and thus no need to update the immediate // dominator of the exit blocks. DT->changeImmediateDominator(LoopExitBlock, EPI.EpilogueIterationCountCheck); // Keep track of bypass blocks, as they feed start values to the induction and // reduction phis in the scalar loop preheader. if (EPI.SCEVSafetyCheck) LoopBypassBlocks.push_back(EPI.SCEVSafetyCheck); if (EPI.MemSafetyCheck) LoopBypassBlocks.push_back(EPI.MemSafetyCheck); LoopBypassBlocks.push_back(EPI.EpilogueIterationCountCheck); // The vec.epilog.iter.check block may contain Phi nodes from inductions or // reductions which merge control-flow from the latch block and the middle // block. Update the incoming values here and move the Phi into the preheader. SmallVector PhisInBlock; for (PHINode &Phi : VecEpilogueIterationCountCheck->phis()) PhisInBlock.push_back(&Phi); for (PHINode *Phi : PhisInBlock) { Phi->moveBefore(LoopVectorPreHeader->getFirstNonPHI()); Phi->replaceIncomingBlockWith( VecEpilogueIterationCountCheck->getSinglePredecessor(), VecEpilogueIterationCountCheck); // If the phi doesn't have an incoming value from the // EpilogueIterationCountCheck, we are done. Otherwise remove the incoming // value and also those from other check blocks. This is needed for // reduction phis only. if (none_of(Phi->blocks(), [&](BasicBlock *IncB) { return EPI.EpilogueIterationCountCheck == IncB; })) continue; Phi->removeIncomingValue(EPI.EpilogueIterationCountCheck); if (EPI.SCEVSafetyCheck) Phi->removeIncomingValue(EPI.SCEVSafetyCheck); if (EPI.MemSafetyCheck) Phi->removeIncomingValue(EPI.MemSafetyCheck); } // Generate a resume induction for the vector epilogue and put it in the // vector epilogue preheader Type *IdxTy = Legal->getWidestInductionType(); PHINode *EPResumeVal = PHINode::Create(IdxTy, 2, "vec.epilog.resume.val"); EPResumeVal->insertBefore(LoopVectorPreHeader->getFirstNonPHIIt()); EPResumeVal->addIncoming(EPI.VectorTripCount, VecEpilogueIterationCountCheck); EPResumeVal->addIncoming(ConstantInt::get(IdxTy, 0), EPI.MainLoopIterationCountCheck); // Generate induction resume values. These variables save the new starting // indexes for the scalar loop. They are used to test if there are any tail // iterations left once the vector loop has completed. // Note that when the vectorized epilogue is skipped due to iteration count // check, then the resume value for the induction variable comes from // the trip count of the main vector loop, hence passing the AdditionalBypass // argument. createInductionResumeValues(ExpandedSCEVs, {VecEpilogueIterationCountCheck, EPI.VectorTripCount} /* AdditionalBypass */); return {completeLoopSkeleton(), EPResumeVal}; } BasicBlock * EpilogueVectorizerEpilogueLoop::emitMinimumVectorEpilogueIterCountCheck( BasicBlock *Bypass, BasicBlock *Insert) { assert(EPI.TripCount && "Expected trip count to have been safed in the first pass."); assert( (!isa(EPI.TripCount) || DT->dominates(cast(EPI.TripCount)->getParent(), Insert)) && "saved trip count does not dominate insertion point."); Value *TC = EPI.TripCount; IRBuilder<> Builder(Insert->getTerminator()); Value *Count = Builder.CreateSub(TC, EPI.VectorTripCount, "n.vec.remaining"); // Generate code to check if the loop's trip count is less than VF * UF of the // vector epilogue loop. auto P = Cost->requiresScalarEpilogue(EPI.EpilogueVF.isVector()) ? ICmpInst::ICMP_ULE : ICmpInst::ICMP_ULT; Value *CheckMinIters = Builder.CreateICmp(P, Count, createStepForVF(Builder, Count->getType(), EPI.EpilogueVF, EPI.EpilogueUF), "min.epilog.iters.check"); BranchInst &BI = *BranchInst::Create(Bypass, LoopVectorPreHeader, CheckMinIters); if (hasBranchWeightMD(*OrigLoop->getLoopLatch()->getTerminator())) { unsigned MainLoopStep = UF * VF.getKnownMinValue(); unsigned EpilogueLoopStep = EPI.EpilogueUF * EPI.EpilogueVF.getKnownMinValue(); // We assume the remaining `Count` is equally distributed in // [0, MainLoopStep) // So the probability for `Count < EpilogueLoopStep` should be // min(MainLoopStep, EpilogueLoopStep) / MainLoopStep unsigned EstimatedSkipCount = std::min(MainLoopStep, EpilogueLoopStep); const uint32_t Weights[] = {EstimatedSkipCount, MainLoopStep - EstimatedSkipCount}; setBranchWeights(BI, Weights); } ReplaceInstWithInst(Insert->getTerminator(), &BI); LoopBypassBlocks.push_back(Insert); return Insert; } void EpilogueVectorizerEpilogueLoop::printDebugTracesAtStart() { LLVM_DEBUG({ dbgs() << "Create Skeleton for epilogue vectorized loop (second pass)\n" << "Epilogue Loop VF:" << EPI.EpilogueVF << ", Epilogue Loop UF:" << EPI.EpilogueUF << "\n"; }); } void EpilogueVectorizerEpilogueLoop::printDebugTracesAtEnd() { DEBUG_WITH_TYPE(VerboseDebug, { dbgs() << "final fn:\n" << *OrigLoop->getHeader()->getParent() << "\n"; }); } bool LoopVectorizationPlanner::getDecisionAndClampRange( const std::function &Predicate, VFRange &Range) { assert(!Range.isEmpty() && "Trying to test an empty VF range."); bool PredicateAtRangeStart = Predicate(Range.Start); for (ElementCount TmpVF : VFRange(Range.Start * 2, Range.End)) if (Predicate(TmpVF) != PredicateAtRangeStart) { Range.End = TmpVF; break; } return PredicateAtRangeStart; } /// Build VPlans for the full range of feasible VF's = {\p MinVF, 2 * \p MinVF, /// 4 * \p MinVF, ..., \p MaxVF} by repeatedly building a VPlan for a sub-range /// of VF's starting at a given VF and extending it as much as possible. Each /// vectorization decision can potentially shorten this sub-range during /// buildVPlan(). void LoopVectorizationPlanner::buildVPlans(ElementCount MinVF, ElementCount MaxVF) { auto MaxVFTimes2 = MaxVF * 2; for (ElementCount VF = MinVF; ElementCount::isKnownLT(VF, MaxVFTimes2);) { VFRange SubRange = {VF, MaxVFTimes2}; VPlans.push_back(buildVPlan(SubRange)); VF = SubRange.End; } } VPValue *VPRecipeBuilder::createEdgeMask(BasicBlock *Src, BasicBlock *Dst, VPlan &Plan) { assert(is_contained(predecessors(Dst), Src) && "Invalid edge"); // Look for cached value. std::pair Edge(Src, Dst); EdgeMaskCacheTy::iterator ECEntryIt = EdgeMaskCache.find(Edge); if (ECEntryIt != EdgeMaskCache.end()) return ECEntryIt->second; VPValue *SrcMask = getBlockInMask(Src); // The terminator has to be a branch inst! BranchInst *BI = dyn_cast(Src->getTerminator()); assert(BI && "Unexpected terminator found"); if (!BI->isConditional() || BI->getSuccessor(0) == BI->getSuccessor(1)) return EdgeMaskCache[Edge] = SrcMask; // If source is an exiting block, we know the exit edge is dynamically dead // in the vector loop, and thus we don't need to restrict the mask. Avoid // adding uses of an otherwise potentially dead instruction. if (OrigLoop->isLoopExiting(Src)) return EdgeMaskCache[Edge] = SrcMask; VPValue *EdgeMask = Plan.getVPValueOrAddLiveIn(BI->getCondition()); assert(EdgeMask && "No Edge Mask found for condition"); if (BI->getSuccessor(0) != Dst) EdgeMask = Builder.createNot(EdgeMask, BI->getDebugLoc()); if (SrcMask) { // Otherwise block in-mask is all-one, no need to AND. // The condition is 'SrcMask && EdgeMask', which is equivalent to // 'select i1 SrcMask, i1 EdgeMask, i1 false'. // The select version does not introduce new UB if SrcMask is false and // EdgeMask is poison. Using 'and' here introduces undefined behavior. VPValue *False = Plan.getVPValueOrAddLiveIn( ConstantInt::getFalse(BI->getCondition()->getType())); EdgeMask = Builder.createSelect(SrcMask, EdgeMask, False, BI->getDebugLoc()); } return EdgeMaskCache[Edge] = EdgeMask; } void VPRecipeBuilder::createHeaderMask(VPlan &Plan) { BasicBlock *Header = OrigLoop->getHeader(); // When not folding the tail, use nullptr to model all-true mask. if (!CM.foldTailByMasking()) { BlockMaskCache[Header] = nullptr; return; } // Introduce the early-exit compare IV <= BTC to form header block mask. // This is used instead of IV < TC because TC may wrap, unlike BTC. Start by // constructing the desired canonical IV in the header block as its first // non-phi instructions. VPBasicBlock *HeaderVPBB = Plan.getVectorLoopRegion()->getEntryBasicBlock(); auto NewInsertionPoint = HeaderVPBB->getFirstNonPhi(); auto *IV = new VPWidenCanonicalIVRecipe(Plan.getCanonicalIV()); HeaderVPBB->insert(IV, NewInsertionPoint); VPBuilder::InsertPointGuard Guard(Builder); Builder.setInsertPoint(HeaderVPBB, NewInsertionPoint); VPValue *BlockMask = nullptr; VPValue *BTC = Plan.getOrCreateBackedgeTakenCount(); BlockMask = Builder.createICmp(CmpInst::ICMP_ULE, IV, BTC); BlockMaskCache[Header] = BlockMask; } VPValue *VPRecipeBuilder::getBlockInMask(BasicBlock *BB) const { // Return the cached value. BlockMaskCacheTy::const_iterator BCEntryIt = BlockMaskCache.find(BB); assert(BCEntryIt != BlockMaskCache.end() && "Trying to access mask for block without one."); return BCEntryIt->second; } void VPRecipeBuilder::createBlockInMask(BasicBlock *BB, VPlan &Plan) { assert(OrigLoop->contains(BB) && "Block is not a part of a loop"); assert(BlockMaskCache.count(BB) == 0 && "Mask for block already computed"); assert(OrigLoop->getHeader() != BB && "Loop header must have cached block mask"); // All-one mask is modelled as no-mask following the convention for masked // load/store/gather/scatter. Initialize BlockMask to no-mask. VPValue *BlockMask = nullptr; // This is the block mask. We OR all incoming edges. for (auto *Predecessor : predecessors(BB)) { VPValue *EdgeMask = createEdgeMask(Predecessor, BB, Plan); if (!EdgeMask) { // Mask of predecessor is all-one so mask of block is too. BlockMaskCache[BB] = EdgeMask; return; } if (!BlockMask) { // BlockMask has its initialized nullptr value. BlockMask = EdgeMask; continue; } BlockMask = Builder.createOr(BlockMask, EdgeMask, {}); } BlockMaskCache[BB] = BlockMask; } VPRecipeBase *VPRecipeBuilder::tryToWidenMemory(Instruction *I, ArrayRef Operands, VFRange &Range, VPlanPtr &Plan) { assert((isa(I) || isa(I)) && "Must be called with either a load or store"); auto willWiden = [&](ElementCount VF) -> bool { LoopVectorizationCostModel::InstWidening Decision = CM.getWideningDecision(I, VF); assert(Decision != LoopVectorizationCostModel::CM_Unknown && "CM decision should be taken at this point."); if (Decision == LoopVectorizationCostModel::CM_Interleave) return true; if (CM.isScalarAfterVectorization(I, VF) || CM.isProfitableToScalarize(I, VF)) return false; return Decision != LoopVectorizationCostModel::CM_Scalarize; }; if (!LoopVectorizationPlanner::getDecisionAndClampRange(willWiden, Range)) return nullptr; VPValue *Mask = nullptr; if (Legal->isMaskRequired(I)) Mask = getBlockInMask(I->getParent()); // Determine if the pointer operand of the access is either consecutive or // reverse consecutive. LoopVectorizationCostModel::InstWidening Decision = CM.getWideningDecision(I, Range.Start); bool Reverse = Decision == LoopVectorizationCostModel::CM_Widen_Reverse; bool Consecutive = Reverse || Decision == LoopVectorizationCostModel::CM_Widen; VPValue *Ptr = isa(I) ? Operands[0] : Operands[1]; if (Consecutive) { auto *GEP = dyn_cast( Ptr->getUnderlyingValue()->stripPointerCasts()); auto *VectorPtr = new VPVectorPointerRecipe( Ptr, getLoadStoreType(I), Reverse, GEP ? GEP->isInBounds() : false, I->getDebugLoc()); Builder.getInsertBlock()->appendRecipe(VectorPtr); Ptr = VectorPtr; } if (LoadInst *Load = dyn_cast(I)) return new VPWidenMemoryInstructionRecipe(*Load, Ptr, Mask, Consecutive, Reverse); StoreInst *Store = cast(I); return new VPWidenMemoryInstructionRecipe(*Store, Ptr, Operands[0], Mask, Consecutive, Reverse); } /// Creates a VPWidenIntOrFpInductionRecpipe for \p Phi. If needed, it will also /// insert a recipe to expand the step for the induction recipe. static VPWidenIntOrFpInductionRecipe * createWidenInductionRecipes(PHINode *Phi, Instruction *PhiOrTrunc, VPValue *Start, const InductionDescriptor &IndDesc, VPlan &Plan, ScalarEvolution &SE, Loop &OrigLoop, VFRange &Range) { assert(IndDesc.getStartValue() == Phi->getIncomingValueForBlock(OrigLoop.getLoopPreheader())); assert(SE.isLoopInvariant(IndDesc.getStep(), &OrigLoop) && "step must be loop invariant"); VPValue *Step = vputils::getOrCreateVPValueForSCEVExpr(Plan, IndDesc.getStep(), SE); if (auto *TruncI = dyn_cast(PhiOrTrunc)) { return new VPWidenIntOrFpInductionRecipe(Phi, Start, Step, IndDesc, TruncI); } assert(isa(PhiOrTrunc) && "must be a phi node here"); return new VPWidenIntOrFpInductionRecipe(Phi, Start, Step, IndDesc); } VPRecipeBase *VPRecipeBuilder::tryToOptimizeInductionPHI( PHINode *Phi, ArrayRef Operands, VPlan &Plan, VFRange &Range) { // Check if this is an integer or fp induction. If so, build the recipe that // produces its scalar and vector values. if (auto *II = Legal->getIntOrFpInductionDescriptor(Phi)) return createWidenInductionRecipes(Phi, Phi, Operands[0], *II, Plan, *PSE.getSE(), *OrigLoop, Range); // Check if this is pointer induction. If so, build the recipe for it. if (auto *II = Legal->getPointerInductionDescriptor(Phi)) { VPValue *Step = vputils::getOrCreateVPValueForSCEVExpr(Plan, II->getStep(), *PSE.getSE()); return new VPWidenPointerInductionRecipe( Phi, Operands[0], Step, *II, LoopVectorizationPlanner::getDecisionAndClampRange( [&](ElementCount VF) { return CM.isScalarAfterVectorization(Phi, VF); }, Range)); } return nullptr; } VPWidenIntOrFpInductionRecipe *VPRecipeBuilder::tryToOptimizeInductionTruncate( TruncInst *I, ArrayRef Operands, VFRange &Range, VPlan &Plan) { // Optimize the special case where the source is a constant integer // induction variable. Notice that we can only optimize the 'trunc' case // because (a) FP conversions lose precision, (b) sext/zext may wrap, and // (c) other casts depend on pointer size. // Determine whether \p K is a truncation based on an induction variable that // can be optimized. auto isOptimizableIVTruncate = [&](Instruction *K) -> std::function { return [=](ElementCount VF) -> bool { return CM.isOptimizableIVTruncate(K, VF); }; }; if (LoopVectorizationPlanner::getDecisionAndClampRange( isOptimizableIVTruncate(I), Range)) { auto *Phi = cast(I->getOperand(0)); const InductionDescriptor &II = *Legal->getIntOrFpInductionDescriptor(Phi); VPValue *Start = Plan.getVPValueOrAddLiveIn(II.getStartValue()); return createWidenInductionRecipes(Phi, I, Start, II, Plan, *PSE.getSE(), *OrigLoop, Range); } return nullptr; } VPRecipeOrVPValueTy VPRecipeBuilder::tryToBlend(PHINode *Phi, ArrayRef Operands, VPlanPtr &Plan) { // If all incoming values are equal, the incoming VPValue can be used directly // instead of creating a new VPBlendRecipe. if (llvm::all_equal(Operands)) return Operands[0]; unsigned NumIncoming = Phi->getNumIncomingValues(); // For in-loop reductions, we do not need to create an additional select. VPValue *InLoopVal = nullptr; for (unsigned In = 0; In < NumIncoming; In++) { PHINode *PhiOp = dyn_cast_or_null(Operands[In]->getUnderlyingValue()); if (PhiOp && CM.isInLoopReduction(PhiOp)) { assert(!InLoopVal && "Found more than one in-loop reduction!"); InLoopVal = Operands[In]; } } assert((!InLoopVal || NumIncoming == 2) && "Found an in-loop reduction for PHI with unexpected number of " "incoming values"); if (InLoopVal) return Operands[Operands[0] == InLoopVal ? 1 : 0]; // We know that all PHIs in non-header blocks are converted into selects, so // we don't have to worry about the insertion order and we can just use the // builder. At this point we generate the predication tree. There may be // duplications since this is a simple recursive scan, but future // optimizations will clean it up. SmallVector OperandsWithMask; for (unsigned In = 0; In < NumIncoming; In++) { VPValue *EdgeMask = createEdgeMask(Phi->getIncomingBlock(In), Phi->getParent(), *Plan); assert((EdgeMask || NumIncoming == 1) && "Multiple predecessors with one having a full mask"); OperandsWithMask.push_back(Operands[In]); if (EdgeMask) OperandsWithMask.push_back(EdgeMask); } return toVPRecipeResult(new VPBlendRecipe(Phi, OperandsWithMask)); } VPWidenCallRecipe *VPRecipeBuilder::tryToWidenCall(CallInst *CI, ArrayRef Operands, VFRange &Range, VPlanPtr &Plan) { bool IsPredicated = LoopVectorizationPlanner::getDecisionAndClampRange( [this, CI](ElementCount VF) { return CM.isScalarWithPredication(CI, VF); }, Range); if (IsPredicated) return nullptr; Intrinsic::ID ID = getVectorIntrinsicIDForCall(CI, TLI); if (ID && (ID == Intrinsic::assume || ID == Intrinsic::lifetime_end || ID == Intrinsic::lifetime_start || ID == Intrinsic::sideeffect || ID == Intrinsic::pseudoprobe || ID == Intrinsic::experimental_noalias_scope_decl)) return nullptr; SmallVector Ops(Operands.take_front(CI->arg_size())); // Is it beneficial to perform intrinsic call compared to lib call? bool ShouldUseVectorIntrinsic = ID && LoopVectorizationPlanner::getDecisionAndClampRange( [&](ElementCount VF) -> bool { return CM.getCallWideningDecision(CI, VF).Kind == LoopVectorizationCostModel::CM_IntrinsicCall; }, Range); if (ShouldUseVectorIntrinsic) return new VPWidenCallRecipe(*CI, make_range(Ops.begin(), Ops.end()), ID, CI->getDebugLoc()); Function *Variant = nullptr; std::optional MaskPos; // Is better to call a vectorized version of the function than to to scalarize // the call? auto ShouldUseVectorCall = LoopVectorizationPlanner::getDecisionAndClampRange( [&](ElementCount VF) -> bool { // The following case may be scalarized depending on the VF. // The flag shows whether we can use a usual Call for vectorized // version of the instruction. // If we've found a variant at a previous VF, then stop looking. A // vectorized variant of a function expects input in a certain shape // -- basically the number of input registers, the number of lanes // per register, and whether there's a mask required. // We store a pointer to the variant in the VPWidenCallRecipe, so // once we have an appropriate variant it's only valid for that VF. // This will force a different vplan to be generated for each VF that // finds a valid variant. if (Variant) return false; LoopVectorizationCostModel::CallWideningDecision Decision = CM.getCallWideningDecision(CI, VF); if (Decision.Kind == LoopVectorizationCostModel::CM_VectorCall) { Variant = Decision.Variant; MaskPos = Decision.MaskPos; return true; } return false; }, Range); if (ShouldUseVectorCall) { if (MaskPos.has_value()) { // We have 2 cases that would require a mask: // 1) The block needs to be predicated, either due to a conditional // in the scalar loop or use of an active lane mask with // tail-folding, and we use the appropriate mask for the block. // 2) No mask is required for the block, but the only available // vector variant at this VF requires a mask, so we synthesize an // all-true mask. VPValue *Mask = nullptr; if (Legal->isMaskRequired(CI)) Mask = getBlockInMask(CI->getParent()); else Mask = Plan->getVPValueOrAddLiveIn(ConstantInt::getTrue( IntegerType::getInt1Ty(Variant->getFunctionType()->getContext()))); Ops.insert(Ops.begin() + *MaskPos, Mask); } return new VPWidenCallRecipe(*CI, make_range(Ops.begin(), Ops.end()), Intrinsic::not_intrinsic, CI->getDebugLoc(), Variant); } return nullptr; } bool VPRecipeBuilder::shouldWiden(Instruction *I, VFRange &Range) const { assert(!isa(I) && !isa(I) && !isa(I) && !isa(I) && "Instruction should have been handled earlier"); // Instruction should be widened, unless it is scalar after vectorization, // scalarization is profitable or it is predicated. auto WillScalarize = [this, I](ElementCount VF) -> bool { return CM.isScalarAfterVectorization(I, VF) || CM.isProfitableToScalarize(I, VF) || CM.isScalarWithPredication(I, VF); }; return !LoopVectorizationPlanner::getDecisionAndClampRange(WillScalarize, Range); } VPRecipeBase *VPRecipeBuilder::tryToWiden(Instruction *I, ArrayRef Operands, VPBasicBlock *VPBB, VPlanPtr &Plan) { switch (I->getOpcode()) { default: return nullptr; case Instruction::SDiv: case Instruction::UDiv: case Instruction::SRem: case Instruction::URem: { // If not provably safe, use a select to form a safe divisor before widening the // div/rem operation itself. Otherwise fall through to general handling below. if (CM.isPredicatedInst(I)) { SmallVector Ops(Operands.begin(), Operands.end()); VPValue *Mask = getBlockInMask(I->getParent()); VPValue *One = Plan->getVPValueOrAddLiveIn( ConstantInt::get(I->getType(), 1u, false)); auto *SafeRHS = new VPInstruction(Instruction::Select, {Mask, Ops[1], One}, I->getDebugLoc()); VPBB->appendRecipe(SafeRHS); Ops[1] = SafeRHS; return new VPWidenRecipe(*I, make_range(Ops.begin(), Ops.end())); } [[fallthrough]]; } case Instruction::Add: case Instruction::And: case Instruction::AShr: case Instruction::FAdd: case Instruction::FCmp: case Instruction::FDiv: case Instruction::FMul: case Instruction::FNeg: case Instruction::FRem: case Instruction::FSub: case Instruction::ICmp: case Instruction::LShr: case Instruction::Mul: case Instruction::Or: case Instruction::Select: case Instruction::Shl: case Instruction::Sub: case Instruction::Xor: case Instruction::Freeze: return new VPWidenRecipe(*I, make_range(Operands.begin(), Operands.end())); }; } void VPRecipeBuilder::fixHeaderPhis() { BasicBlock *OrigLatch = OrigLoop->getLoopLatch(); for (VPHeaderPHIRecipe *R : PhisToFix) { auto *PN = cast(R->getUnderlyingValue()); VPRecipeBase *IncR = getRecipe(cast(PN->getIncomingValueForBlock(OrigLatch))); R->addOperand(IncR->getVPSingleValue()); } } VPRecipeOrVPValueTy VPRecipeBuilder::handleReplication(Instruction *I, VFRange &Range, VPlan &Plan) { bool IsUniform = LoopVectorizationPlanner::getDecisionAndClampRange( [&](ElementCount VF) { return CM.isUniformAfterVectorization(I, VF); }, Range); bool IsPredicated = CM.isPredicatedInst(I); // Even if the instruction is not marked as uniform, there are certain // intrinsic calls that can be effectively treated as such, so we check for // them here. Conservatively, we only do this for scalable vectors, since // for fixed-width VFs we can always fall back on full scalarization. if (!IsUniform && Range.Start.isScalable() && isa(I)) { switch (cast(I)->getIntrinsicID()) { case Intrinsic::assume: case Intrinsic::lifetime_start: case Intrinsic::lifetime_end: // For scalable vectors if one of the operands is variant then we still // want to mark as uniform, which will generate one instruction for just // the first lane of the vector. We can't scalarize the call in the same // way as for fixed-width vectors because we don't know how many lanes // there are. // // The reasons for doing it this way for scalable vectors are: // 1. For the assume intrinsic generating the instruction for the first // lane is still be better than not generating any at all. For // example, the input may be a splat across all lanes. // 2. For the lifetime start/end intrinsics the pointer operand only // does anything useful when the input comes from a stack object, // which suggests it should always be uniform. For non-stack objects // the effect is to poison the object, which still allows us to // remove the call. IsUniform = true; break; default: break; } } VPValue *BlockInMask = nullptr; if (!IsPredicated) { // Finalize the recipe for Instr, first if it is not predicated. LLVM_DEBUG(dbgs() << "LV: Scalarizing:" << *I << "\n"); } else { LLVM_DEBUG(dbgs() << "LV: Scalarizing and predicating:" << *I << "\n"); // Instructions marked for predication are replicated and a mask operand is // added initially. Masked replicate recipes will later be placed under an // if-then construct to prevent side-effects. Generate recipes to compute // the block mask for this region. BlockInMask = getBlockInMask(I->getParent()); } auto *Recipe = new VPReplicateRecipe(I, Plan.mapToVPValues(I->operands()), IsUniform, BlockInMask); return toVPRecipeResult(Recipe); } VPRecipeOrVPValueTy VPRecipeBuilder::tryToCreateWidenRecipe(Instruction *Instr, ArrayRef Operands, VFRange &Range, VPBasicBlock *VPBB, VPlanPtr &Plan) { // First, check for specific widening recipes that deal with inductions, Phi // nodes, calls and memory operations. VPRecipeBase *Recipe; if (auto Phi = dyn_cast(Instr)) { if (Phi->getParent() != OrigLoop->getHeader()) return tryToBlend(Phi, Operands, Plan); // Always record recipes for header phis. Later first-order recurrence phis // can have earlier phis as incoming values. recordRecipeOf(Phi); if ((Recipe = tryToOptimizeInductionPHI(Phi, Operands, *Plan, Range))) return toVPRecipeResult(Recipe); VPHeaderPHIRecipe *PhiRecipe = nullptr; assert((Legal->isReductionVariable(Phi) || Legal->isFixedOrderRecurrence(Phi)) && "can only widen reductions and fixed-order recurrences here"); VPValue *StartV = Operands[0]; if (Legal->isReductionVariable(Phi)) { const RecurrenceDescriptor &RdxDesc = Legal->getReductionVars().find(Phi)->second; assert(RdxDesc.getRecurrenceStartValue() == Phi->getIncomingValueForBlock(OrigLoop->getLoopPreheader())); PhiRecipe = new VPReductionPHIRecipe(Phi, RdxDesc, *StartV, CM.isInLoopReduction(Phi), CM.useOrderedReductions(RdxDesc)); } else { // TODO: Currently fixed-order recurrences are modeled as chains of // first-order recurrences. If there are no users of the intermediate // recurrences in the chain, the fixed order recurrence should be modeled // directly, enabling more efficient codegen. PhiRecipe = new VPFirstOrderRecurrencePHIRecipe(Phi, *StartV); } // Record the incoming value from the backedge, so we can add the incoming // value from the backedge after all recipes have been created. auto *Inc = cast( Phi->getIncomingValueForBlock(OrigLoop->getLoopLatch())); auto RecipeIter = Ingredient2Recipe.find(Inc); if (RecipeIter == Ingredient2Recipe.end()) recordRecipeOf(Inc); PhisToFix.push_back(PhiRecipe); return toVPRecipeResult(PhiRecipe); } if (isa(Instr) && (Recipe = tryToOptimizeInductionTruncate(cast(Instr), Operands, Range, *Plan))) return toVPRecipeResult(Recipe); // All widen recipes below deal only with VF > 1. if (LoopVectorizationPlanner::getDecisionAndClampRange( [&](ElementCount VF) { return VF.isScalar(); }, Range)) return nullptr; if (auto *CI = dyn_cast(Instr)) return toVPRecipeResult(tryToWidenCall(CI, Operands, Range, Plan)); if (isa(Instr) || isa(Instr)) return toVPRecipeResult(tryToWidenMemory(Instr, Operands, Range, Plan)); if (!shouldWiden(Instr, Range)) return nullptr; if (auto GEP = dyn_cast(Instr)) return toVPRecipeResult(new VPWidenGEPRecipe( GEP, make_range(Operands.begin(), Operands.end()))); if (auto *SI = dyn_cast(Instr)) { return toVPRecipeResult(new VPWidenSelectRecipe( *SI, make_range(Operands.begin(), Operands.end()))); } if (auto *CI = dyn_cast(Instr)) { return toVPRecipeResult(new VPWidenCastRecipe(CI->getOpcode(), Operands[0], CI->getType(), *CI)); } return toVPRecipeResult(tryToWiden(Instr, Operands, VPBB, Plan)); } void LoopVectorizationPlanner::buildVPlansWithVPRecipes(ElementCount MinVF, ElementCount MaxVF) { assert(OrigLoop->isInnermost() && "Inner loop expected."); auto MaxVFTimes2 = MaxVF * 2; for (ElementCount VF = MinVF; ElementCount::isKnownLT(VF, MaxVFTimes2);) { VFRange SubRange = {VF, MaxVFTimes2}; if (auto Plan = tryToBuildVPlanWithVPRecipes(SubRange)) { // Now optimize the initial VPlan. if (!Plan->hasVF(ElementCount::getFixed(1))) VPlanTransforms::truncateToMinimalBitwidths( *Plan, CM.getMinimalBitwidths(), PSE.getSE()->getContext()); VPlanTransforms::optimize(*Plan, *PSE.getSE()); assert(VPlanVerifier::verifyPlanIsValid(*Plan) && "VPlan is invalid"); VPlans.push_back(std::move(Plan)); } VF = SubRange.End; } } // Add the necessary canonical IV and branch recipes required to control the // loop. static void addCanonicalIVRecipes(VPlan &Plan, Type *IdxTy, bool HasNUW, DebugLoc DL) { Value *StartIdx = ConstantInt::get(IdxTy, 0); auto *StartV = Plan.getVPValueOrAddLiveIn(StartIdx); // Add a VPCanonicalIVPHIRecipe starting at 0 to the header. auto *CanonicalIVPHI = new VPCanonicalIVPHIRecipe(StartV, DL); VPRegionBlock *TopRegion = Plan.getVectorLoopRegion(); VPBasicBlock *Header = TopRegion->getEntryBasicBlock(); Header->insert(CanonicalIVPHI, Header->begin()); // Add a CanonicalIVIncrement{NUW} VPInstruction to increment the scalar // IV by VF * UF. auto *CanonicalIVIncrement = new VPInstruction(Instruction::Add, {CanonicalIVPHI, &Plan.getVFxUF()}, {HasNUW, false}, DL, "index.next"); CanonicalIVPHI->addOperand(CanonicalIVIncrement); VPBasicBlock *EB = TopRegion->getExitingBasicBlock(); EB->appendRecipe(CanonicalIVIncrement); // Add the BranchOnCount VPInstruction to the latch. VPInstruction *BranchBack = new VPInstruction(VPInstruction::BranchOnCount, {CanonicalIVIncrement, &Plan.getVectorTripCount()}, DL); EB->appendRecipe(BranchBack); } // Add exit values to \p Plan. VPLiveOuts are added for each LCSSA phi in the // original exit block. static void addUsersInExitBlock(VPBasicBlock *HeaderVPBB, Loop *OrigLoop, VPlan &Plan) { BasicBlock *ExitBB = OrigLoop->getUniqueExitBlock(); BasicBlock *ExitingBB = OrigLoop->getExitingBlock(); // Only handle single-exit loops with unique exit blocks for now. if (!ExitBB || !ExitBB->getSinglePredecessor() || !ExitingBB) return; // Introduce VPUsers modeling the exit values. for (PHINode &ExitPhi : ExitBB->phis()) { Value *IncomingValue = ExitPhi.getIncomingValueForBlock(ExitingBB); VPValue *V = Plan.getVPValueOrAddLiveIn(IncomingValue); Plan.addLiveOut(&ExitPhi, V); } } VPlanPtr LoopVectorizationPlanner::tryToBuildVPlanWithVPRecipes(VFRange &Range) { SmallPtrSet *, 1> InterleaveGroups; VPRecipeBuilder RecipeBuilder(OrigLoop, TLI, Legal, CM, PSE, Builder); // --------------------------------------------------------------------------- // Pre-construction: record ingredients whose recipes we'll need to further // process after constructing the initial VPlan. // --------------------------------------------------------------------------- // For each interleave group which is relevant for this (possibly trimmed) // Range, add it to the set of groups to be later applied to the VPlan and add // placeholders for its members' Recipes which we'll be replacing with a // single VPInterleaveRecipe. for (InterleaveGroup *IG : IAI.getInterleaveGroups()) { auto applyIG = [IG, this](ElementCount VF) -> bool { bool Result = (VF.isVector() && // Query is illegal for VF == 1 CM.getWideningDecision(IG->getInsertPos(), VF) == LoopVectorizationCostModel::CM_Interleave); // For scalable vectors, the only interleave factor currently supported // is 2 since we require the (de)interleave2 intrinsics instead of // shufflevectors. assert((!Result || !VF.isScalable() || IG->getFactor() == 2) && "Unsupported interleave factor for scalable vectors"); return Result; }; if (!getDecisionAndClampRange(applyIG, Range)) continue; InterleaveGroups.insert(IG); for (unsigned i = 0; i < IG->getFactor(); i++) if (Instruction *Member = IG->getMember(i)) RecipeBuilder.recordRecipeOf(Member); }; // --------------------------------------------------------------------------- // Build initial VPlan: Scan the body of the loop in a topological order to // visit each basic block after having visited its predecessor basic blocks. // --------------------------------------------------------------------------- // Create initial VPlan skeleton, having a basic block for the pre-header // which contains SCEV expansions that need to happen before the CFG is // modified; a basic block for the vector pre-header, followed by a region for // the vector loop, followed by the middle basic block. The skeleton vector // loop region contains a header and latch basic blocks. VPlanPtr Plan = VPlan::createInitialVPlan( createTripCountSCEV(Legal->getWidestInductionType(), PSE, OrigLoop), *PSE.getSE()); VPBasicBlock *HeaderVPBB = new VPBasicBlock("vector.body"); VPBasicBlock *LatchVPBB = new VPBasicBlock("vector.latch"); VPBlockUtils::insertBlockAfter(LatchVPBB, HeaderVPBB); Plan->getVectorLoopRegion()->setEntry(HeaderVPBB); Plan->getVectorLoopRegion()->setExiting(LatchVPBB); // Don't use getDecisionAndClampRange here, because we don't know the UF // so this function is better to be conservative, rather than to split // it up into different VPlans. // TODO: Consider using getDecisionAndClampRange here to split up VPlans. bool IVUpdateMayOverflow = false; for (ElementCount VF : Range) IVUpdateMayOverflow |= !isIndvarOverflowCheckKnownFalse(&CM, VF); DebugLoc DL = getDebugLocFromInstOrOperands(Legal->getPrimaryInduction()); TailFoldingStyle Style = CM.getTailFoldingStyle(IVUpdateMayOverflow); // When not folding the tail, we know that the induction increment will not // overflow. bool HasNUW = Style == TailFoldingStyle::None; addCanonicalIVRecipes(*Plan, Legal->getWidestInductionType(), HasNUW, DL); // Scan the body of the loop in a topological order to visit each basic block // after having visited its predecessor basic blocks. LoopBlocksDFS DFS(OrigLoop); DFS.perform(LI); VPBasicBlock *VPBB = HeaderVPBB; bool NeedsMasks = CM.foldTailByMasking() || any_of(OrigLoop->blocks(), [this](BasicBlock *BB) { return Legal->blockNeedsPredication(BB); }); for (BasicBlock *BB : make_range(DFS.beginRPO(), DFS.endRPO())) { // Relevant instructions from basic block BB will be grouped into VPRecipe // ingredients and fill a new VPBasicBlock. if (VPBB != HeaderVPBB) VPBB->setName(BB->getName()); Builder.setInsertPoint(VPBB); if (VPBB == HeaderVPBB) RecipeBuilder.createHeaderMask(*Plan); else if (NeedsMasks) RecipeBuilder.createBlockInMask(BB, *Plan); // Introduce each ingredient into VPlan. // TODO: Model and preserve debug intrinsics in VPlan. for (Instruction &I : drop_end(BB->instructionsWithoutDebug(false))) { Instruction *Instr = &I; SmallVector Operands; auto *Phi = dyn_cast(Instr); if (Phi && Phi->getParent() == OrigLoop->getHeader()) { Operands.push_back(Plan->getVPValueOrAddLiveIn( Phi->getIncomingValueForBlock(OrigLoop->getLoopPreheader()))); } else { auto OpRange = Plan->mapToVPValues(Instr->operands()); Operands = {OpRange.begin(), OpRange.end()}; } // Invariant stores inside loop will be deleted and a single store // with the final reduction value will be added to the exit block StoreInst *SI; if ((SI = dyn_cast(&I)) && Legal->isInvariantAddressOfReduction(SI->getPointerOperand())) continue; auto RecipeOrValue = RecipeBuilder.tryToCreateWidenRecipe( Instr, Operands, Range, VPBB, Plan); if (!RecipeOrValue) RecipeOrValue = RecipeBuilder.handleReplication(Instr, Range, *Plan); // If Instr can be simplified to an existing VPValue, use it. if (isa(RecipeOrValue)) { auto *VPV = cast(RecipeOrValue); Plan->addVPValue(Instr, VPV); // If the re-used value is a recipe, register the recipe for the // instruction, in case the recipe for Instr needs to be recorded. if (VPRecipeBase *R = VPV->getDefiningRecipe()) RecipeBuilder.setRecipe(Instr, R); continue; } // Otherwise, add the new recipe. VPRecipeBase *Recipe = cast(RecipeOrValue); for (auto *Def : Recipe->definedValues()) { auto *UV = Def->getUnderlyingValue(); Plan->addVPValue(UV, Def); } RecipeBuilder.setRecipe(Instr, Recipe); if (isa(Recipe)) { // VPHeaderPHIRecipes must be kept in the phi section of HeaderVPBB. In // the following cases, VPHeaderPHIRecipes may be created after non-phi // recipes and need to be moved to the phi section of HeaderVPBB: // * tail-folding (non-phi recipes computing the header mask are // introduced earlier than regular header phi recipes, and should appear // after them) // * Optimizing truncates to VPWidenIntOrFpInductionRecipe. assert((HeaderVPBB->getFirstNonPhi() == VPBB->end() || CM.foldTailByMasking() || isa(Instr)) && "unexpected recipe needs moving"); Recipe->insertBefore(*HeaderVPBB, HeaderVPBB->getFirstNonPhi()); } else VPBB->appendRecipe(Recipe); } VPBlockUtils::insertBlockAfter(new VPBasicBlock(), VPBB); VPBB = cast(VPBB->getSingleSuccessor()); } // After here, VPBB should not be used. VPBB = nullptr; if (CM.requiresScalarEpilogue(Range)) { // No edge from the middle block to the unique exit block has been inserted // and there is nothing to fix from vector loop; phis should have incoming // from scalar loop only. } else addUsersInExitBlock(HeaderVPBB, OrigLoop, *Plan); assert(isa(Plan->getVectorLoopRegion()) && !Plan->getVectorLoopRegion()->getEntryBasicBlock()->empty() && "entry block must be set to a VPRegionBlock having a non-empty entry " "VPBasicBlock"); RecipeBuilder.fixHeaderPhis(); // --------------------------------------------------------------------------- // Transform initial VPlan: Apply previously taken decisions, in order, to // bring the VPlan to its final state. // --------------------------------------------------------------------------- // Adjust the recipes for any inloop reductions. adjustRecipesForReductions(LatchVPBB, Plan, RecipeBuilder, Range.Start); // Interleave memory: for each Interleave Group we marked earlier as relevant // for this VPlan, replace the Recipes widening its memory instructions with a // single VPInterleaveRecipe at its insertion point. for (const auto *IG : InterleaveGroups) { auto *Recipe = cast( RecipeBuilder.getRecipe(IG->getInsertPos())); SmallVector StoredValues; for (unsigned i = 0; i < IG->getFactor(); ++i) if (auto *SI = dyn_cast_or_null(IG->getMember(i))) { auto *StoreR = cast(RecipeBuilder.getRecipe(SI)); StoredValues.push_back(StoreR->getStoredValue()); } bool NeedsMaskForGaps = IG->requiresScalarEpilogue() && !CM.isScalarEpilogueAllowed(); auto *VPIG = new VPInterleaveRecipe(IG, Recipe->getAddr(), StoredValues, Recipe->getMask(), NeedsMaskForGaps); VPIG->insertBefore(Recipe); unsigned J = 0; for (unsigned i = 0; i < IG->getFactor(); ++i) if (Instruction *Member = IG->getMember(i)) { VPRecipeBase *MemberR = RecipeBuilder.getRecipe(Member); if (!Member->getType()->isVoidTy()) { VPValue *OriginalV = MemberR->getVPSingleValue(); OriginalV->replaceAllUsesWith(VPIG->getVPValue(J)); J++; } MemberR->eraseFromParent(); } } for (ElementCount VF : Range) Plan->addVF(VF); Plan->setName("Initial VPlan"); // Replace VPValues for known constant strides guaranteed by predicate scalar // evolution. for (auto [_, Stride] : Legal->getLAI()->getSymbolicStrides()) { auto *StrideV = cast(Stride)->getValue(); auto *ScevStride = dyn_cast(PSE.getSCEV(StrideV)); // Only handle constant strides for now. if (!ScevStride) continue; Constant *CI = ConstantInt::get(Stride->getType(), ScevStride->getAPInt()); auto *ConstVPV = Plan->getVPValueOrAddLiveIn(CI); // The versioned value may not be used in the loop directly, so just add a // new live-in in those cases. Plan->getVPValueOrAddLiveIn(StrideV)->replaceAllUsesWith(ConstVPV); } // From this point onwards, VPlan-to-VPlan transformations may change the plan // in ways that accessing values using original IR values is incorrect. Plan->disableValue2VPValue(); // Sink users of fixed-order recurrence past the recipe defining the previous // value and introduce FirstOrderRecurrenceSplice VPInstructions. if (!VPlanTransforms::adjustFixedOrderRecurrences(*Plan, Builder)) return nullptr; if (useActiveLaneMask(Style)) { // TODO: Move checks to VPlanTransforms::addActiveLaneMask once // TailFoldingStyle is visible there. bool ForControlFlow = useActiveLaneMaskForControlFlow(Style); bool WithoutRuntimeCheck = Style == TailFoldingStyle::DataAndControlFlowWithoutRuntimeCheck; VPlanTransforms::addActiveLaneMask(*Plan, ForControlFlow, WithoutRuntimeCheck); } return Plan; } VPlanPtr LoopVectorizationPlanner::buildVPlan(VFRange &Range) { // Outer loop handling: They may require CFG and instruction level // transformations before even evaluating whether vectorization is profitable. // Since we cannot modify the incoming IR, we need to build VPlan upfront in // the vectorization pipeline. assert(!OrigLoop->isInnermost()); assert(EnableVPlanNativePath && "VPlan-native path is not enabled."); // Create new empty VPlan auto Plan = VPlan::createInitialVPlan( createTripCountSCEV(Legal->getWidestInductionType(), PSE, OrigLoop), *PSE.getSE()); // Build hierarchical CFG VPlanHCFGBuilder HCFGBuilder(OrigLoop, LI, *Plan); HCFGBuilder.buildHierarchicalCFG(); for (ElementCount VF : Range) Plan->addVF(VF); VPlanTransforms::VPInstructionsToVPRecipes( Plan, [this](PHINode *P) { return Legal->getIntOrFpInductionDescriptor(P); }, *PSE.getSE(), *TLI); // Remove the existing terminator of the exiting block of the top-most region. // A BranchOnCount will be added instead when adding the canonical IV recipes. auto *Term = Plan->getVectorLoopRegion()->getExitingBasicBlock()->getTerminator(); Term->eraseFromParent(); // Tail folding is not supported for outer loops, so the induction increment // is guaranteed to not wrap. bool HasNUW = true; addCanonicalIVRecipes(*Plan, Legal->getWidestInductionType(), HasNUW, DebugLoc()); return Plan; } // Adjust the recipes for reductions. For in-loop reductions the chain of // instructions leading from the loop exit instr to the phi need to be converted // to reductions, with one operand being vector and the other being the scalar // reduction chain. For other reductions, a select is introduced between the phi // and live-out recipes when folding the tail. // // A ComputeReductionResult recipe is added to the middle block, also for // in-loop reductions which compute their result in-loop, because generating // the subsequent bc.merge.rdx phi is driven by ComputeReductionResult recipes. void LoopVectorizationPlanner::adjustRecipesForReductions( VPBasicBlock *LatchVPBB, VPlanPtr &Plan, VPRecipeBuilder &RecipeBuilder, ElementCount MinVF) { VPRegionBlock *VectorLoopRegion = Plan->getVectorLoopRegion(); VPBasicBlock *Header = VectorLoopRegion->getEntryBasicBlock(); // Gather all VPReductionPHIRecipe and sort them so that Intermediate stores // sank outside of the loop would keep the same order as they had in the // original loop. SmallVector ReductionPHIList; for (VPRecipeBase &R : Header->phis()) { if (auto *ReductionPhi = dyn_cast(&R)) ReductionPHIList.emplace_back(ReductionPhi); } bool HasIntermediateStore = false; stable_sort(ReductionPHIList, [this, &HasIntermediateStore](const VPReductionPHIRecipe *R1, const VPReductionPHIRecipe *R2) { auto *IS1 = R1->getRecurrenceDescriptor().IntermediateStore; auto *IS2 = R2->getRecurrenceDescriptor().IntermediateStore; HasIntermediateStore |= IS1 || IS2; // If neither of the recipes has an intermediate store, keep the // order the same. if (!IS1 && !IS2) return false; // If only one of the recipes has an intermediate store, then // move it towards the beginning of the list. if (IS1 && !IS2) return true; if (!IS1 && IS2) return false; // If both recipes have an intermediate store, then the recipe // with the later store should be processed earlier. So it // should go to the beginning of the list. return DT->dominates(IS2, IS1); }); if (HasIntermediateStore && ReductionPHIList.size() > 1) for (VPRecipeBase *R : ReductionPHIList) R->moveBefore(*Header, Header->getFirstNonPhi()); for (VPRecipeBase &R : Header->phis()) { auto *PhiR = dyn_cast(&R); if (!PhiR || !PhiR->isInLoop() || (MinVF.isScalar() && !PhiR->isOrdered())) continue; const RecurrenceDescriptor &RdxDesc = PhiR->getRecurrenceDescriptor(); RecurKind Kind = RdxDesc.getRecurrenceKind(); assert(!RecurrenceDescriptor::isAnyOfRecurrenceKind(Kind) && "AnyOf reductions are not allowed for in-loop reductions"); // Collect the chain of "link" recipes for the reduction starting at PhiR. SetVector Worklist; Worklist.insert(PhiR); for (unsigned I = 0; I != Worklist.size(); ++I) { VPSingleDefRecipe *Cur = Worklist[I]; for (VPUser *U : Cur->users()) { auto *UserRecipe = dyn_cast(U); if (!UserRecipe) { assert(isa(U) && "U must either be a VPSingleDef or VPLiveOut"); continue; } Worklist.insert(UserRecipe); } } // Visit operation "Links" along the reduction chain top-down starting from // the phi until LoopExitValue. We keep track of the previous item // (PreviousLink) to tell which of the two operands of a Link will remain // scalar and which will be reduced. For minmax by select(cmp), Link will be // the select instructions. VPSingleDefRecipe *PreviousLink = PhiR; // Aka Worklist[0]. for (VPSingleDefRecipe *CurrentLink : Worklist.getArrayRef().drop_front()) { Instruction *CurrentLinkI = CurrentLink->getUnderlyingInstr(); // Index of the first operand which holds a non-mask vector operand. unsigned IndexOfFirstOperand; // Recognize a call to the llvm.fmuladd intrinsic. bool IsFMulAdd = (Kind == RecurKind::FMulAdd); VPValue *VecOp; VPBasicBlock *LinkVPBB = CurrentLink->getParent(); if (IsFMulAdd) { assert( RecurrenceDescriptor::isFMulAddIntrinsic(CurrentLinkI) && "Expected instruction to be a call to the llvm.fmuladd intrinsic"); assert(((MinVF.isScalar() && isa(CurrentLink)) || isa(CurrentLink)) && CurrentLink->getOperand(2) == PreviousLink && "expected a call where the previous link is the added operand"); // If the instruction is a call to the llvm.fmuladd intrinsic then we // need to create an fmul recipe (multiplying the first two operands of // the fmuladd together) to use as the vector operand for the fadd // reduction. VPInstruction *FMulRecipe = new VPInstruction( Instruction::FMul, {CurrentLink->getOperand(0), CurrentLink->getOperand(1)}, CurrentLinkI->getFastMathFlags()); LinkVPBB->insert(FMulRecipe, CurrentLink->getIterator()); VecOp = FMulRecipe; } else { if (RecurrenceDescriptor::isMinMaxRecurrenceKind(Kind)) { if (isa(CurrentLink)) { assert(isa(CurrentLinkI) && "need to have the compare of the select"); continue; } assert(isa(CurrentLink) && "must be a select recipe"); IndexOfFirstOperand = 1; } else { assert((MinVF.isScalar() || isa(CurrentLink)) && "Expected to replace a VPWidenSC"); IndexOfFirstOperand = 0; } // Note that for non-commutable operands (cmp-selects), the semantics of // the cmp-select are captured in the recurrence kind. unsigned VecOpId = CurrentLink->getOperand(IndexOfFirstOperand) == PreviousLink ? IndexOfFirstOperand + 1 : IndexOfFirstOperand; VecOp = CurrentLink->getOperand(VecOpId); assert(VecOp != PreviousLink && CurrentLink->getOperand(CurrentLink->getNumOperands() - 1 - (VecOpId - IndexOfFirstOperand)) == PreviousLink && "PreviousLink must be the operand other than VecOp"); } BasicBlock *BB = CurrentLinkI->getParent(); VPValue *CondOp = nullptr; if (CM.blockNeedsPredicationForAnyReason(BB)) { VPBuilder::InsertPointGuard Guard(Builder); Builder.setInsertPoint(CurrentLink); CondOp = RecipeBuilder.getBlockInMask(BB); } VPReductionRecipe *RedRecipe = new VPReductionRecipe( RdxDesc, CurrentLinkI, PreviousLink, VecOp, CondOp); // Append the recipe to the end of the VPBasicBlock because we need to // ensure that it comes after all of it's inputs, including CondOp. // Note that this transformation may leave over dead recipes (including // CurrentLink), which will be cleaned by a later VPlan transform. LinkVPBB->appendRecipe(RedRecipe); CurrentLink->replaceAllUsesWith(RedRecipe); PreviousLink = RedRecipe; } } Builder.setInsertPoint(&*LatchVPBB->begin()); for (VPRecipeBase &R : Plan->getVectorLoopRegion()->getEntryBasicBlock()->phis()) { VPReductionPHIRecipe *PhiR = dyn_cast(&R); if (!PhiR) continue; const RecurrenceDescriptor &RdxDesc = PhiR->getRecurrenceDescriptor(); // If tail is folded by masking, introduce selects between the phi // and the live-out instruction of each reduction, at the beginning of the // dedicated latch block. auto *OrigExitingVPV = PhiR->getBackedgeValue(); auto *NewExitingVPV = PhiR->getBackedgeValue(); if (!PhiR->isInLoop() && CM.foldTailByMasking()) { VPValue *Cond = RecipeBuilder.getBlockInMask(OrigLoop->getHeader()); assert(OrigExitingVPV->getDefiningRecipe()->getParent() != LatchVPBB && "reduction recipe must be defined before latch"); Type *PhiTy = PhiR->getOperand(0)->getLiveInIRValue()->getType(); std::optional FMFs = PhiTy->isFloatingPointTy() ? std::make_optional(RdxDesc.getFastMathFlags()) : std::nullopt; NewExitingVPV = Builder.createSelect(Cond, OrigExitingVPV, PhiR, {}, "", FMFs); OrigExitingVPV->replaceUsesWithIf(NewExitingVPV, [](VPUser &U, unsigned) { return isa(&U) && cast(&U)->getOpcode() == VPInstruction::ComputeReductionResult; }); if (PreferPredicatedReductionSelect || TTI.preferPredicatedReductionSelect( PhiR->getRecurrenceDescriptor().getOpcode(), PhiTy, TargetTransformInfo::ReductionFlags())) PhiR->setOperand(1, NewExitingVPV); } // If the vector reduction can be performed in a smaller type, we truncate // then extend the loop exit value to enable InstCombine to evaluate the // entire expression in the smaller type. Type *PhiTy = PhiR->getStartValue()->getLiveInIRValue()->getType(); if (MinVF.isVector() && PhiTy != RdxDesc.getRecurrenceType()) { assert(!PhiR->isInLoop() && "Unexpected truncated inloop reduction!"); Type *RdxTy = RdxDesc.getRecurrenceType(); auto *Trunc = new VPWidenCastRecipe(Instruction::Trunc, NewExitingVPV, RdxTy); auto *Extnd = RdxDesc.isSigned() ? new VPWidenCastRecipe(Instruction::SExt, Trunc, PhiTy) : new VPWidenCastRecipe(Instruction::ZExt, Trunc, PhiTy); Trunc->insertAfter(NewExitingVPV->getDefiningRecipe()); Extnd->insertAfter(Trunc); if (PhiR->getOperand(1) == NewExitingVPV) PhiR->setOperand(1, Extnd->getVPSingleValue()); NewExitingVPV = Extnd; } // We want code in the middle block to appear to execute on the location of // the scalar loop's latch terminator because: (a) it is all compiler // generated, (b) these instructions are always executed after evaluating // the latch conditional branch, and (c) other passes may add new // predecessors which terminate on this line. This is the easiest way to // ensure we don't accidentally cause an extra step back into the loop while // debugging. DebugLoc ExitDL = OrigLoop->getLoopLatch()->getTerminator()->getDebugLoc(); // TODO: At the moment ComputeReductionResult also drives creation of the // bc.merge.rdx phi nodes, hence it needs to be created unconditionally here // even for in-loop reductions, until the reduction resume value handling is // also modeled in VPlan. auto *FinalReductionResult = new VPInstruction( VPInstruction::ComputeReductionResult, {PhiR, NewExitingVPV}, ExitDL); cast(VectorLoopRegion->getSingleSuccessor()) ->appendRecipe(FinalReductionResult); OrigExitingVPV->replaceUsesWithIf( FinalReductionResult, [](VPUser &User, unsigned) { return isa(&User); }); } VPlanTransforms::clearReductionWrapFlags(*Plan); } #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP) void VPInterleaveRecipe::print(raw_ostream &O, const Twine &Indent, VPSlotTracker &SlotTracker) const { O << Indent << "INTERLEAVE-GROUP with factor " << IG->getFactor() << " at "; IG->getInsertPos()->printAsOperand(O, false); O << ", "; getAddr()->printAsOperand(O, SlotTracker); VPValue *Mask = getMask(); if (Mask) { O << ", "; Mask->printAsOperand(O, SlotTracker); } unsigned OpIdx = 0; for (unsigned i = 0; i < IG->getFactor(); ++i) { if (!IG->getMember(i)) continue; if (getNumStoreOperands() > 0) { O << "\n" << Indent << " store "; getOperand(1 + OpIdx)->printAsOperand(O, SlotTracker); O << " to index " << i; } else { O << "\n" << Indent << " "; getVPValue(OpIdx)->printAsOperand(O, SlotTracker); O << " = load from index " << i; } ++OpIdx; } } #endif void VPWidenPointerInductionRecipe::execute(VPTransformState &State) { assert(IndDesc.getKind() == InductionDescriptor::IK_PtrInduction && "Not a pointer induction according to InductionDescriptor!"); assert(cast(getUnderlyingInstr())->getType()->isPointerTy() && "Unexpected type."); auto *IVR = getParent()->getPlan()->getCanonicalIV(); PHINode *CanonicalIV = cast(State.get(IVR, 0)); if (onlyScalarsGenerated(State.VF)) { // This is the normalized GEP that starts counting at zero. Value *PtrInd = State.Builder.CreateSExtOrTrunc( CanonicalIV, IndDesc.getStep()->getType()); // Determine the number of scalars we need to generate for each unroll // iteration. If the instruction is uniform, we only need to generate the // first lane. Otherwise, we generate all VF values. bool IsUniform = vputils::onlyFirstLaneUsed(this); assert((IsUniform || !State.VF.isScalable()) && "Cannot scalarize a scalable VF"); unsigned Lanes = IsUniform ? 1 : State.VF.getFixedValue(); for (unsigned Part = 0; Part < State.UF; ++Part) { Value *PartStart = createStepForVF(State.Builder, PtrInd->getType(), State.VF, Part); for (unsigned Lane = 0; Lane < Lanes; ++Lane) { Value *Idx = State.Builder.CreateAdd( PartStart, ConstantInt::get(PtrInd->getType(), Lane)); Value *GlobalIdx = State.Builder.CreateAdd(PtrInd, Idx); Value *Step = State.get(getOperand(1), VPIteration(Part, Lane)); Value *SclrGep = emitTransformedIndex( State.Builder, GlobalIdx, IndDesc.getStartValue(), Step, IndDesc.getKind(), IndDesc.getInductionBinOp()); SclrGep->setName("next.gep"); State.set(this, SclrGep, VPIteration(Part, Lane)); } } return; } Type *PhiType = IndDesc.getStep()->getType(); // Build a pointer phi Value *ScalarStartValue = getStartValue()->getLiveInIRValue(); Type *ScStValueType = ScalarStartValue->getType(); PHINode *NewPointerPhi = PHINode::Create(ScStValueType, 2, "pointer.phi", CanonicalIV); BasicBlock *VectorPH = State.CFG.getPreheaderBBFor(this); NewPointerPhi->addIncoming(ScalarStartValue, VectorPH); // A pointer induction, performed by using a gep Instruction *InductionLoc = &*State.Builder.GetInsertPoint(); Value *ScalarStepValue = State.get(getOperand(1), VPIteration(0, 0)); Value *RuntimeVF = getRuntimeVF(State.Builder, PhiType, State.VF); Value *NumUnrolledElems = State.Builder.CreateMul(RuntimeVF, ConstantInt::get(PhiType, State.UF)); Value *InductionGEP = GetElementPtrInst::Create( State.Builder.getInt8Ty(), NewPointerPhi, State.Builder.CreateMul(ScalarStepValue, NumUnrolledElems), "ptr.ind", InductionLoc); // Add induction update using an incorrect block temporarily. The phi node // will be fixed after VPlan execution. Note that at this point the latch // block cannot be used, as it does not exist yet. // TODO: Model increment value in VPlan, by turning the recipe into a // multi-def and a subclass of VPHeaderPHIRecipe. NewPointerPhi->addIncoming(InductionGEP, VectorPH); // Create UF many actual address geps that use the pointer // phi as base and a vectorized version of the step value // () as offset. for (unsigned Part = 0; Part < State.UF; ++Part) { Type *VecPhiType = VectorType::get(PhiType, State.VF); Value *StartOffsetScalar = State.Builder.CreateMul(RuntimeVF, ConstantInt::get(PhiType, Part)); Value *StartOffset = State.Builder.CreateVectorSplat(State.VF, StartOffsetScalar); // Create a vector of consecutive numbers from zero to VF. StartOffset = State.Builder.CreateAdd( StartOffset, State.Builder.CreateStepVector(VecPhiType)); assert(ScalarStepValue == State.get(getOperand(1), VPIteration(Part, 0)) && "scalar step must be the same across all parts"); Value *GEP = State.Builder.CreateGEP( State.Builder.getInt8Ty(), NewPointerPhi, State.Builder.CreateMul( StartOffset, State.Builder.CreateVectorSplat(State.VF, ScalarStepValue), "vector.gep")); State.set(this, GEP, Part); } } void VPDerivedIVRecipe::execute(VPTransformState &State) { assert(!State.Instance && "VPDerivedIVRecipe being replicated."); // Fast-math-flags propagate from the original induction instruction. IRBuilder<>::FastMathFlagGuard FMFG(State.Builder); if (FPBinOp) State.Builder.setFastMathFlags(FPBinOp->getFastMathFlags()); Value *Step = State.get(getStepValue(), VPIteration(0, 0)); Value *CanonicalIV = State.get(getCanonicalIV(), VPIteration(0, 0)); Value *DerivedIV = emitTransformedIndex( State.Builder, CanonicalIV, getStartValue()->getLiveInIRValue(), Step, Kind, cast_if_present(FPBinOp)); DerivedIV->setName("offset.idx"); if (TruncResultTy) { assert(TruncResultTy != DerivedIV->getType() && Step->getType()->isIntegerTy() && "Truncation requires an integer step"); DerivedIV = State.Builder.CreateTrunc(DerivedIV, TruncResultTy); } assert(DerivedIV != CanonicalIV && "IV didn't need transforming?"); State.set(this, DerivedIV, VPIteration(0, 0)); } void VPInterleaveRecipe::execute(VPTransformState &State) { assert(!State.Instance && "Interleave group being replicated."); State.ILV->vectorizeInterleaveGroup(IG, definedValues(), State, getAddr(), getStoredValues(), getMask(), NeedsMaskForGaps); } void VPReductionRecipe::execute(VPTransformState &State) { assert(!State.Instance && "Reduction being replicated."); Value *PrevInChain = State.get(getChainOp(), 0); RecurKind Kind = RdxDesc.getRecurrenceKind(); bool IsOrdered = State.ILV->useOrderedReductions(RdxDesc); // Propagate the fast-math flags carried by the underlying instruction. IRBuilderBase::FastMathFlagGuard FMFGuard(State.Builder); State.Builder.setFastMathFlags(RdxDesc.getFastMathFlags()); for (unsigned Part = 0; Part < State.UF; ++Part) { Value *NewVecOp = State.get(getVecOp(), Part); if (VPValue *Cond = getCondOp()) { Value *NewCond = State.VF.isVector() ? State.get(Cond, Part) : State.get(Cond, {Part, 0}); VectorType *VecTy = dyn_cast(NewVecOp->getType()); Type *ElementTy = VecTy ? VecTy->getElementType() : NewVecOp->getType(); Value *Iden = RdxDesc.getRecurrenceIdentity(Kind, ElementTy, RdxDesc.getFastMathFlags()); if (State.VF.isVector()) { Iden = State.Builder.CreateVectorSplat(VecTy->getElementCount(), Iden); } Value *Select = State.Builder.CreateSelect(NewCond, NewVecOp, Iden); NewVecOp = Select; } Value *NewRed; Value *NextInChain; if (IsOrdered) { if (State.VF.isVector()) NewRed = createOrderedReduction(State.Builder, RdxDesc, NewVecOp, PrevInChain); else NewRed = State.Builder.CreateBinOp( (Instruction::BinaryOps)RdxDesc.getOpcode(Kind), PrevInChain, NewVecOp); PrevInChain = NewRed; } else { PrevInChain = State.get(getChainOp(), Part); NewRed = createTargetReduction(State.Builder, RdxDesc, NewVecOp); } if (RecurrenceDescriptor::isMinMaxRecurrenceKind(Kind)) { NextInChain = createMinMaxOp(State.Builder, RdxDesc.getRecurrenceKind(), NewRed, PrevInChain); } else if (IsOrdered) NextInChain = NewRed; else NextInChain = State.Builder.CreateBinOp( (Instruction::BinaryOps)RdxDesc.getOpcode(Kind), NewRed, PrevInChain); State.set(this, NextInChain, Part); } } void VPReplicateRecipe::execute(VPTransformState &State) { Instruction *UI = getUnderlyingInstr(); if (State.Instance) { // Generate a single instance. assert(!State.VF.isScalable() && "Can't scalarize a scalable vector"); State.ILV->scalarizeInstruction(UI, this, *State.Instance, State); // Insert scalar instance packing it into a vector. if (State.VF.isVector() && shouldPack()) { // If we're constructing lane 0, initialize to start from poison. if (State.Instance->Lane.isFirstLane()) { assert(!State.VF.isScalable() && "VF is assumed to be non scalable."); Value *Poison = PoisonValue::get( VectorType::get(UI->getType(), State.VF)); State.set(this, Poison, State.Instance->Part); } State.packScalarIntoVectorValue(this, *State.Instance); } return; } if (IsUniform) { // If the recipe is uniform across all parts (instead of just per VF), only // generate a single instance. if ((isa(UI) || isa(UI)) && all_of(operands(), [](VPValue *Op) { return Op->isDefinedOutsideVectorRegions(); })) { State.ILV->scalarizeInstruction(UI, this, VPIteration(0, 0), State); if (user_begin() != user_end()) { for (unsigned Part = 1; Part < State.UF; ++Part) State.set(this, State.get(this, VPIteration(0, 0)), VPIteration(Part, 0)); } return; } // Uniform within VL means we need to generate lane 0 only for each // unrolled copy. for (unsigned Part = 0; Part < State.UF; ++Part) State.ILV->scalarizeInstruction(UI, this, VPIteration(Part, 0), State); return; } // A store of a loop varying value to a uniform address only needs the last // copy of the store. if (isa(UI) && vputils::isUniformAfterVectorization(getOperand(1))) { auto Lane = VPLane::getLastLaneForVF(State.VF); State.ILV->scalarizeInstruction(UI, this, VPIteration(State.UF - 1, Lane), State); return; } // Generate scalar instances for all VF lanes of all UF parts. assert(!State.VF.isScalable() && "Can't scalarize a scalable vector"); const unsigned EndLane = State.VF.getKnownMinValue(); for (unsigned Part = 0; Part < State.UF; ++Part) for (unsigned Lane = 0; Lane < EndLane; ++Lane) State.ILV->scalarizeInstruction(UI, this, VPIteration(Part, Lane), State); } void VPWidenMemoryInstructionRecipe::execute(VPTransformState &State) { VPValue *StoredValue = isStore() ? getStoredValue() : nullptr; // Attempt to issue a wide load. LoadInst *LI = dyn_cast(&Ingredient); StoreInst *SI = dyn_cast(&Ingredient); assert((LI || SI) && "Invalid Load/Store instruction"); assert((!SI || StoredValue) && "No stored value provided for widened store"); assert((!LI || !StoredValue) && "Stored value provided for widened load"); Type *ScalarDataTy = getLoadStoreType(&Ingredient); auto *DataTy = VectorType::get(ScalarDataTy, State.VF); const Align Alignment = getLoadStoreAlignment(&Ingredient); bool CreateGatherScatter = !isConsecutive(); auto &Builder = State.Builder; InnerLoopVectorizer::VectorParts BlockInMaskParts(State.UF); bool isMaskRequired = getMask(); if (isMaskRequired) { // Mask reversal is only needed for non-all-one (null) masks, as reverse of // a null all-one mask is a null mask. for (unsigned Part = 0; Part < State.UF; ++Part) { Value *Mask = State.get(getMask(), Part); if (isReverse()) Mask = Builder.CreateVectorReverse(Mask, "reverse"); BlockInMaskParts[Part] = Mask; } } // Handle Stores: if (SI) { State.setDebugLocFrom(SI->getDebugLoc()); for (unsigned Part = 0; Part < State.UF; ++Part) { Instruction *NewSI = nullptr; Value *StoredVal = State.get(StoredValue, Part); if (CreateGatherScatter) { Value *MaskPart = isMaskRequired ? BlockInMaskParts[Part] : nullptr; Value *VectorGep = State.get(getAddr(), Part); NewSI = Builder.CreateMaskedScatter(StoredVal, VectorGep, Alignment, MaskPart); } else { if (isReverse()) { // If we store to reverse consecutive memory locations, then we need // to reverse the order of elements in the stored value. StoredVal = Builder.CreateVectorReverse(StoredVal, "reverse"); // We don't want to update the value in the map as it might be used in // another expression. So don't call resetVectorValue(StoredVal). } auto *VecPtr = State.get(getAddr(), Part); if (isMaskRequired) NewSI = Builder.CreateMaskedStore(StoredVal, VecPtr, Alignment, BlockInMaskParts[Part]); else NewSI = Builder.CreateAlignedStore(StoredVal, VecPtr, Alignment); } State.addMetadata(NewSI, SI); } return; } // Handle loads. assert(LI && "Must have a load instruction"); State.setDebugLocFrom(LI->getDebugLoc()); for (unsigned Part = 0; Part < State.UF; ++Part) { Value *NewLI; if (CreateGatherScatter) { Value *MaskPart = isMaskRequired ? BlockInMaskParts[Part] : nullptr; Value *VectorGep = State.get(getAddr(), Part); NewLI = Builder.CreateMaskedGather(DataTy, VectorGep, Alignment, MaskPart, nullptr, "wide.masked.gather"); State.addMetadata(NewLI, LI); } else { auto *VecPtr = State.get(getAddr(), Part); if (isMaskRequired) NewLI = Builder.CreateMaskedLoad( DataTy, VecPtr, Alignment, BlockInMaskParts[Part], PoisonValue::get(DataTy), "wide.masked.load"); else NewLI = Builder.CreateAlignedLoad(DataTy, VecPtr, Alignment, "wide.load"); // Add metadata to the load, but setVectorValue to the reverse shuffle. State.addMetadata(NewLI, LI); if (Reverse) NewLI = Builder.CreateVectorReverse(NewLI, "reverse"); } State.set(getVPSingleValue(), NewLI, Part); } } // Determine how to lower the scalar epilogue, which depends on 1) optimising // for minimum code-size, 2) predicate compiler options, 3) loop hints forcing // predication, and 4) a TTI hook that analyses whether the loop is suitable // for predication. static ScalarEpilogueLowering getScalarEpilogueLowering( Function *F, Loop *L, LoopVectorizeHints &Hints, ProfileSummaryInfo *PSI, BlockFrequencyInfo *BFI, TargetTransformInfo *TTI, TargetLibraryInfo *TLI, LoopVectorizationLegality &LVL, InterleavedAccessInfo *IAI) { // 1) OptSize takes precedence over all other options, i.e. if this is set, // don't look at hints or options, and don't request a scalar epilogue. // (For PGSO, as shouldOptimizeForSize isn't currently accessible from // LoopAccessInfo (due to code dependency and not being able to reliably get // PSI/BFI from a loop analysis under NPM), we cannot suppress the collection // of strides in LoopAccessInfo::analyzeLoop() and vectorize without // versioning when the vectorization is forced, unlike hasOptSize. So revert // back to the old way and vectorize with versioning when forced. See D81345.) if (F->hasOptSize() || (llvm::shouldOptimizeForSize(L->getHeader(), PSI, BFI, PGSOQueryType::IRPass) && Hints.getForce() != LoopVectorizeHints::FK_Enabled)) return CM_ScalarEpilogueNotAllowedOptSize; // 2) If set, obey the directives if (PreferPredicateOverEpilogue.getNumOccurrences()) { switch (PreferPredicateOverEpilogue) { case PreferPredicateTy::ScalarEpilogue: return CM_ScalarEpilogueAllowed; case PreferPredicateTy::PredicateElseScalarEpilogue: return CM_ScalarEpilogueNotNeededUsePredicate; case PreferPredicateTy::PredicateOrDontVectorize: return CM_ScalarEpilogueNotAllowedUsePredicate; }; } // 3) If set, obey the hints switch (Hints.getPredicate()) { case LoopVectorizeHints::FK_Enabled: return CM_ScalarEpilogueNotNeededUsePredicate; case LoopVectorizeHints::FK_Disabled: return CM_ScalarEpilogueAllowed; }; // 4) if the TTI hook indicates this is profitable, request predication. TailFoldingInfo TFI(TLI, &LVL, IAI); if (TTI->preferPredicateOverEpilogue(&TFI)) return CM_ScalarEpilogueNotNeededUsePredicate; return CM_ScalarEpilogueAllowed; } // Process the loop in the VPlan-native vectorization path. This path builds // VPlan upfront in the vectorization pipeline, which allows to apply // VPlan-to-VPlan transformations from the very beginning without modifying the // input LLVM IR. static bool processLoopInVPlanNativePath( Loop *L, PredicatedScalarEvolution &PSE, LoopInfo *LI, DominatorTree *DT, LoopVectorizationLegality *LVL, TargetTransformInfo *TTI, TargetLibraryInfo *TLI, DemandedBits *DB, AssumptionCache *AC, OptimizationRemarkEmitter *ORE, BlockFrequencyInfo *BFI, ProfileSummaryInfo *PSI, LoopVectorizeHints &Hints, LoopVectorizationRequirements &Requirements) { if (isa(PSE.getBackedgeTakenCount())) { LLVM_DEBUG(dbgs() << "LV: cannot compute the outer-loop trip count\n"); return false; } assert(EnableVPlanNativePath && "VPlan-native path is disabled."); Function *F = L->getHeader()->getParent(); InterleavedAccessInfo IAI(PSE, L, DT, LI, LVL->getLAI()); ScalarEpilogueLowering SEL = getScalarEpilogueLowering(F, L, Hints, PSI, BFI, TTI, TLI, *LVL, &IAI); LoopVectorizationCostModel CM(SEL, L, PSE, LI, LVL, *TTI, TLI, DB, AC, ORE, F, &Hints, IAI); // Use the planner for outer loop vectorization. // TODO: CM is not used at this point inside the planner. Turn CM into an // optional argument if we don't need it in the future. LoopVectorizationPlanner LVP(L, LI, DT, TLI, *TTI, LVL, CM, IAI, PSE, Hints, ORE); // Get user vectorization factor. ElementCount UserVF = Hints.getWidth(); CM.collectElementTypesForWidening(); // Plan how to best vectorize, return the best VF and its cost. const VectorizationFactor VF = LVP.planInVPlanNativePath(UserVF); // If we are stress testing VPlan builds, do not attempt to generate vector // code. Masked vector code generation support will follow soon. // Also, do not attempt to vectorize if no vector code will be produced. if (VPlanBuildStressTest || VectorizationFactor::Disabled() == VF) return false; VPlan &BestPlan = LVP.getBestPlanFor(VF.Width); { bool AddBranchWeights = hasBranchWeightMD(*L->getLoopLatch()->getTerminator()); GeneratedRTChecks Checks(*PSE.getSE(), DT, LI, TTI, F->getParent()->getDataLayout(), AddBranchWeights); InnerLoopVectorizer LB(L, PSE, LI, DT, TLI, TTI, AC, ORE, VF.Width, VF.Width, 1, LVL, &CM, BFI, PSI, Checks); LLVM_DEBUG(dbgs() << "Vectorizing outer loop in \"" << L->getHeader()->getParent()->getName() << "\"\n"); LVP.executePlan(VF.Width, 1, BestPlan, LB, DT, false); } reportVectorization(ORE, L, VF, 1); // Mark the loop as already vectorized to avoid vectorizing again. Hints.setAlreadyVectorized(); assert(!verifyFunction(*L->getHeader()->getParent(), &dbgs())); return true; } // Emit a remark if there are stores to floats that required a floating point // extension. If the vectorized loop was generated with floating point there // will be a performance penalty from the conversion overhead and the change in // the vector width. static void checkMixedPrecision(Loop *L, OptimizationRemarkEmitter *ORE) { SmallVector Worklist; for (BasicBlock *BB : L->getBlocks()) { for (Instruction &Inst : *BB) { if (auto *S = dyn_cast(&Inst)) { if (S->getValueOperand()->getType()->isFloatTy()) Worklist.push_back(S); } } } // Traverse the floating point stores upwards searching, for floating point // conversions. SmallPtrSet Visited; SmallPtrSet EmittedRemark; while (!Worklist.empty()) { auto *I = Worklist.pop_back_val(); if (!L->contains(I)) continue; if (!Visited.insert(I).second) continue; // Emit a remark if the floating point store required a floating // point conversion. // TODO: More work could be done to identify the root cause such as a // constant or a function return type and point the user to it. if (isa(I) && EmittedRemark.insert(I).second) ORE->emit([&]() { return OptimizationRemarkAnalysis(LV_NAME, "VectorMixedPrecision", I->getDebugLoc(), L->getHeader()) << "floating point conversion changes vector width. " << "Mixed floating point precision requires an up/down " << "cast that will negatively impact performance."; }); for (Use &Op : I->operands()) if (auto *OpI = dyn_cast(Op)) Worklist.push_back(OpI); } } static bool areRuntimeChecksProfitable(GeneratedRTChecks &Checks, VectorizationFactor &VF, std::optional VScale, Loop *L, ScalarEvolution &SE, ScalarEpilogueLowering SEL) { InstructionCost CheckCost = Checks.getCost(); if (!CheckCost.isValid()) return false; // When interleaving only scalar and vector cost will be equal, which in turn // would lead to a divide by 0. Fall back to hard threshold. if (VF.Width.isScalar()) { if (CheckCost > VectorizeMemoryCheckThreshold) { LLVM_DEBUG( dbgs() << "LV: Interleaving only is not profitable due to runtime checks\n"); return false; } return true; } // The scalar cost should only be 0 when vectorizing with a user specified VF/IC. In those cases, runtime checks should always be generated. double ScalarC = *VF.ScalarCost.getValue(); if (ScalarC == 0) return true; // First, compute the minimum iteration count required so that the vector // loop outperforms the scalar loop. // The total cost of the scalar loop is // ScalarC * TC // where // * TC is the actual trip count of the loop. // * ScalarC is the cost of a single scalar iteration. // // The total cost of the vector loop is // RtC + VecC * (TC / VF) + EpiC // where // * RtC is the cost of the generated runtime checks // * VecC is the cost of a single vector iteration. // * TC is the actual trip count of the loop // * VF is the vectorization factor // * EpiCost is the cost of the generated epilogue, including the cost // of the remaining scalar operations. // // Vectorization is profitable once the total vector cost is less than the // total scalar cost: // RtC + VecC * (TC / VF) + EpiC < ScalarC * TC // // Now we can compute the minimum required trip count TC as // (RtC + EpiC) / (ScalarC - (VecC / VF)) < TC // // For now we assume the epilogue cost EpiC = 0 for simplicity. Note that // the computations are performed on doubles, not integers and the result // is rounded up, hence we get an upper estimate of the TC. unsigned IntVF = VF.Width.getKnownMinValue(); if (VF.Width.isScalable()) { unsigned AssumedMinimumVscale = 1; if (VScale) AssumedMinimumVscale = *VScale; IntVF *= AssumedMinimumVscale; } double VecCOverVF = double(*VF.Cost.getValue()) / IntVF; double RtC = *CheckCost.getValue(); double MinTC1 = RtC / (ScalarC - VecCOverVF); // Second, compute a minimum iteration count so that the cost of the // runtime checks is only a fraction of the total scalar loop cost. This // adds a loop-dependent bound on the overhead incurred if the runtime // checks fail. In case the runtime checks fail, the cost is RtC + ScalarC // * TC. To bound the runtime check to be a fraction 1/X of the scalar // cost, compute // RtC < ScalarC * TC * (1 / X) ==> RtC * X / ScalarC < TC double MinTC2 = RtC * 10 / ScalarC; // Now pick the larger minimum. If it is not a multiple of VF and a scalar // epilogue is allowed, choose the next closest multiple of VF. This should // partly compensate for ignoring the epilogue cost. uint64_t MinTC = std::ceil(std::max(MinTC1, MinTC2)); if (SEL == CM_ScalarEpilogueAllowed) MinTC = alignTo(MinTC, IntVF); VF.MinProfitableTripCount = ElementCount::getFixed(MinTC); LLVM_DEBUG( dbgs() << "LV: Minimum required TC for runtime checks to be profitable:" << VF.MinProfitableTripCount << "\n"); // Skip vectorization if the expected trip count is less than the minimum // required trip count. if (auto ExpectedTC = getSmallBestKnownTC(SE, L)) { if (ElementCount::isKnownLT(ElementCount::getFixed(*ExpectedTC), VF.MinProfitableTripCount)) { LLVM_DEBUG(dbgs() << "LV: Vectorization is not beneficial: expected " "trip count < minimum profitable VF (" << *ExpectedTC << " < " << VF.MinProfitableTripCount << ")\n"); return false; } } return true; } LoopVectorizePass::LoopVectorizePass(LoopVectorizeOptions Opts) : InterleaveOnlyWhenForced(Opts.InterleaveOnlyWhenForced || !EnableLoopInterleaving), VectorizeOnlyWhenForced(Opts.VectorizeOnlyWhenForced || !EnableLoopVectorization) {} bool LoopVectorizePass::processLoop(Loop *L) { assert((EnableVPlanNativePath || L->isInnermost()) && "VPlan-native path is not enabled. Only process inner loops."); #ifndef NDEBUG const std::string DebugLocStr = getDebugLocString(L); #endif /* NDEBUG */ LLVM_DEBUG(dbgs() << "\nLV: Checking a loop in '" << L->getHeader()->getParent()->getName() << "' from " << DebugLocStr << "\n"); LoopVectorizeHints Hints(L, InterleaveOnlyWhenForced, *ORE, TTI); LLVM_DEBUG( dbgs() << "LV: Loop hints:" << " force=" << (Hints.getForce() == LoopVectorizeHints::FK_Disabled ? "disabled" : (Hints.getForce() == LoopVectorizeHints::FK_Enabled ? "enabled" : "?")) << " width=" << Hints.getWidth() << " interleave=" << Hints.getInterleave() << "\n"); // Function containing loop Function *F = L->getHeader()->getParent(); // Looking at the diagnostic output is the only way to determine if a loop // was vectorized (other than looking at the IR or machine code), so it // is important to generate an optimization remark for each loop. Most of // these messages are generated as OptimizationRemarkAnalysis. Remarks // generated as OptimizationRemark and OptimizationRemarkMissed are // less verbose reporting vectorized loops and unvectorized loops that may // benefit from vectorization, respectively. if (!Hints.allowVectorization(F, L, VectorizeOnlyWhenForced)) { LLVM_DEBUG(dbgs() << "LV: Loop hints prevent vectorization.\n"); return false; } PredicatedScalarEvolution PSE(*SE, *L); // Check if it is legal to vectorize the loop. LoopVectorizationRequirements Requirements; LoopVectorizationLegality LVL(L, PSE, DT, TTI, TLI, F, *LAIs, LI, ORE, &Requirements, &Hints, DB, AC, BFI, PSI); if (!LVL.canVectorize(EnableVPlanNativePath)) { LLVM_DEBUG(dbgs() << "LV: Not vectorizing: Cannot prove legality.\n"); Hints.emitRemarkWithHints(); return false; } // Entrance to the VPlan-native vectorization path. Outer loops are processed // here. They may require CFG and instruction level transformations before // even evaluating whether vectorization is profitable. Since we cannot modify // the incoming IR, we need to build VPlan upfront in the vectorization // pipeline. if (!L->isInnermost()) return processLoopInVPlanNativePath(L, PSE, LI, DT, &LVL, TTI, TLI, DB, AC, ORE, BFI, PSI, Hints, Requirements); assert(L->isInnermost() && "Inner loop expected."); InterleavedAccessInfo IAI(PSE, L, DT, LI, LVL.getLAI()); bool UseInterleaved = TTI->enableInterleavedAccessVectorization(); // If an override option has been passed in for interleaved accesses, use it. if (EnableInterleavedMemAccesses.getNumOccurrences() > 0) UseInterleaved = EnableInterleavedMemAccesses; // Analyze interleaved memory accesses. if (UseInterleaved) IAI.analyzeInterleaving(useMaskedInterleavedAccesses(*TTI)); // Check the function attributes and profiles to find out if this function // should be optimized for size. ScalarEpilogueLowering SEL = getScalarEpilogueLowering(F, L, Hints, PSI, BFI, TTI, TLI, LVL, &IAI); // Check the loop for a trip count threshold: vectorize loops with a tiny trip // count by optimizing for size, to minimize overheads. auto ExpectedTC = getSmallBestKnownTC(*SE, L); if (ExpectedTC && *ExpectedTC < TinyTripCountVectorThreshold) { LLVM_DEBUG(dbgs() << "LV: Found a loop with a very small trip count. " << "This loop is worth vectorizing only if no scalar " << "iteration overheads are incurred."); if (Hints.getForce() == LoopVectorizeHints::FK_Enabled) LLVM_DEBUG(dbgs() << " But vectorizing was explicitly forced.\n"); else { if (*ExpectedTC > TTI->getMinTripCountTailFoldingThreshold()) { LLVM_DEBUG(dbgs() << "\n"); // Predicate tail-folded loops are efficient even when the loop // iteration count is low. However, setting the epilogue policy to // `CM_ScalarEpilogueNotAllowedLowTripLoop` prevents vectorizing loops // with runtime checks. It's more effective to let // `areRuntimeChecksProfitable` determine if vectorization is beneficial // for the loop. if (SEL != CM_ScalarEpilogueNotNeededUsePredicate) SEL = CM_ScalarEpilogueNotAllowedLowTripLoop; } else { LLVM_DEBUG(dbgs() << " But the target considers the trip count too " "small to consider vectorizing.\n"); reportVectorizationFailure( "The trip count is below the minial threshold value.", "loop trip count is too low, avoiding vectorization", "LowTripCount", ORE, L); Hints.emitRemarkWithHints(); return false; } } } // Check the function attributes to see if implicit floats or vectors are // allowed. if (F->hasFnAttribute(Attribute::NoImplicitFloat)) { reportVectorizationFailure( "Can't vectorize when the NoImplicitFloat attribute is used", "loop not vectorized due to NoImplicitFloat attribute", "NoImplicitFloat", ORE, L); Hints.emitRemarkWithHints(); return false; } // Check if the target supports potentially unsafe FP vectorization. // FIXME: Add a check for the type of safety issue (denormal, signaling) // for the target we're vectorizing for, to make sure none of the // additional fp-math flags can help. if (Hints.isPotentiallyUnsafe() && TTI->isFPVectorizationPotentiallyUnsafe()) { reportVectorizationFailure( "Potentially unsafe FP op prevents vectorization", "loop not vectorized due to unsafe FP support.", "UnsafeFP", ORE, L); Hints.emitRemarkWithHints(); return false; } bool AllowOrderedReductions; // If the flag is set, use that instead and override the TTI behaviour. if (ForceOrderedReductions.getNumOccurrences() > 0) AllowOrderedReductions = ForceOrderedReductions; else AllowOrderedReductions = TTI->enableOrderedReductions(); if (!LVL.canVectorizeFPMath(AllowOrderedReductions)) { ORE->emit([&]() { auto *ExactFPMathInst = Requirements.getExactFPInst(); return OptimizationRemarkAnalysisFPCommute(DEBUG_TYPE, "CantReorderFPOps", ExactFPMathInst->getDebugLoc(), ExactFPMathInst->getParent()) << "loop not vectorized: cannot prove it is safe to reorder " "floating-point operations"; }); LLVM_DEBUG(dbgs() << "LV: loop not vectorized: cannot prove it is safe to " "reorder floating-point operations\n"); Hints.emitRemarkWithHints(); return false; } // Use the cost model. LoopVectorizationCostModel CM(SEL, L, PSE, LI, &LVL, *TTI, TLI, DB, AC, ORE, F, &Hints, IAI); // Use the planner for vectorization. LoopVectorizationPlanner LVP(L, LI, DT, TLI, *TTI, &LVL, CM, IAI, PSE, Hints, ORE); // Get user vectorization factor and interleave count. ElementCount UserVF = Hints.getWidth(); unsigned UserIC = Hints.getInterleave(); // Plan how to best vectorize, return the best VF and its cost. std::optional MaybeVF = LVP.plan(UserVF, UserIC); VectorizationFactor VF = VectorizationFactor::Disabled(); unsigned IC = 1; bool AddBranchWeights = hasBranchWeightMD(*L->getLoopLatch()->getTerminator()); GeneratedRTChecks Checks(*PSE.getSE(), DT, LI, TTI, F->getParent()->getDataLayout(), AddBranchWeights); if (MaybeVF) { VF = *MaybeVF; // Select the interleave count. IC = CM.selectInterleaveCount(VF.Width, VF.Cost); unsigned SelectedIC = std::max(IC, UserIC); // Optimistically generate runtime checks if they are needed. Drop them if // they turn out to not be profitable. if (VF.Width.isVector() || SelectedIC > 1) Checks.Create(L, *LVL.getLAI(), PSE.getPredicate(), VF.Width, SelectedIC); // Check if it is profitable to vectorize with runtime checks. bool ForceVectorization = Hints.getForce() == LoopVectorizeHints::FK_Enabled; if (!ForceVectorization && !areRuntimeChecksProfitable(Checks, VF, getVScaleForTuning(L, *TTI), L, *PSE.getSE(), SEL)) { ORE->emit([&]() { return OptimizationRemarkAnalysisAliasing( DEBUG_TYPE, "CantReorderMemOps", L->getStartLoc(), L->getHeader()) << "loop not vectorized: cannot prove it is safe to reorder " "memory operations"; }); LLVM_DEBUG(dbgs() << "LV: Too many memory checks needed.\n"); Hints.emitRemarkWithHints(); return false; } } // Identify the diagnostic messages that should be produced. std::pair VecDiagMsg, IntDiagMsg; bool VectorizeLoop = true, InterleaveLoop = true; if (VF.Width.isScalar()) { LLVM_DEBUG(dbgs() << "LV: Vectorization is possible but not beneficial.\n"); VecDiagMsg = std::make_pair( "VectorizationNotBeneficial", "the cost-model indicates that vectorization is not beneficial"); VectorizeLoop = false; } if (!MaybeVF && UserIC > 1) { // Tell the user interleaving was avoided up-front, despite being explicitly // requested. LLVM_DEBUG(dbgs() << "LV: Ignoring UserIC, because vectorization and " "interleaving should be avoided up front\n"); IntDiagMsg = std::make_pair( "InterleavingAvoided", "Ignoring UserIC, because interleaving was avoided up front"); InterleaveLoop = false; } else if (IC == 1 && UserIC <= 1) { // Tell the user interleaving is not beneficial. LLVM_DEBUG(dbgs() << "LV: Interleaving is not beneficial.\n"); IntDiagMsg = std::make_pair( "InterleavingNotBeneficial", "the cost-model indicates that interleaving is not beneficial"); InterleaveLoop = false; if (UserIC == 1) { IntDiagMsg.first = "InterleavingNotBeneficialAndDisabled"; IntDiagMsg.second += " and is explicitly disabled or interleave count is set to 1"; } } else if (IC > 1 && UserIC == 1) { // Tell the user interleaving is beneficial, but it explicitly disabled. LLVM_DEBUG( dbgs() << "LV: Interleaving is beneficial but is explicitly disabled."); IntDiagMsg = std::make_pair( "InterleavingBeneficialButDisabled", "the cost-model indicates that interleaving is beneficial " "but is explicitly disabled or interleave count is set to 1"); InterleaveLoop = false; } // Override IC if user provided an interleave count. IC = UserIC > 0 ? UserIC : IC; // Emit diagnostic messages, if any. const char *VAPassName = Hints.vectorizeAnalysisPassName(); if (!VectorizeLoop && !InterleaveLoop) { // Do not vectorize or interleaving the loop. ORE->emit([&]() { return OptimizationRemarkMissed(VAPassName, VecDiagMsg.first, L->getStartLoc(), L->getHeader()) << VecDiagMsg.second; }); ORE->emit([&]() { return OptimizationRemarkMissed(LV_NAME, IntDiagMsg.first, L->getStartLoc(), L->getHeader()) << IntDiagMsg.second; }); return false; } else if (!VectorizeLoop && InterleaveLoop) { LLVM_DEBUG(dbgs() << "LV: Interleave Count is " << IC << '\n'); ORE->emit([&]() { return OptimizationRemarkAnalysis(VAPassName, VecDiagMsg.first, L->getStartLoc(), L->getHeader()) << VecDiagMsg.second; }); } else if (VectorizeLoop && !InterleaveLoop) { LLVM_DEBUG(dbgs() << "LV: Found a vectorizable loop (" << VF.Width << ") in " << DebugLocStr << '\n'); ORE->emit([&]() { return OptimizationRemarkAnalysis(LV_NAME, IntDiagMsg.first, L->getStartLoc(), L->getHeader()) << IntDiagMsg.second; }); } else if (VectorizeLoop && InterleaveLoop) { LLVM_DEBUG(dbgs() << "LV: Found a vectorizable loop (" << VF.Width << ") in " << DebugLocStr << '\n'); LLVM_DEBUG(dbgs() << "LV: Interleave Count is " << IC << '\n'); } bool DisableRuntimeUnroll = false; MDNode *OrigLoopID = L->getLoopID(); { using namespace ore; if (!VectorizeLoop) { assert(IC > 1 && "interleave count should not be 1 or 0"); // If we decided that it is not legal to vectorize the loop, then // interleave it. InnerLoopUnroller Unroller(L, PSE, LI, DT, TLI, TTI, AC, ORE, IC, &LVL, &CM, BFI, PSI, Checks); VPlan &BestPlan = LVP.getBestPlanFor(VF.Width); LVP.executePlan(VF.Width, IC, BestPlan, Unroller, DT, false); ORE->emit([&]() { return OptimizationRemark(LV_NAME, "Interleaved", L->getStartLoc(), L->getHeader()) << "interleaved loop (interleaved count: " << NV("InterleaveCount", IC) << ")"; }); } else { // If we decided that it is *legal* to vectorize the loop, then do it. // Consider vectorizing the epilogue too if it's profitable. VectorizationFactor EpilogueVF = LVP.selectEpilogueVectorizationFactor(VF.Width, IC); if (EpilogueVF.Width.isVector()) { // The first pass vectorizes the main loop and creates a scalar epilogue // to be vectorized by executing the plan (potentially with a different // factor) again shortly afterwards. EpilogueLoopVectorizationInfo EPI(VF.Width, IC, EpilogueVF.Width, 1); EpilogueVectorizerMainLoop MainILV(L, PSE, LI, DT, TLI, TTI, AC, ORE, EPI, &LVL, &CM, BFI, PSI, Checks); VPlan &BestMainPlan = LVP.getBestPlanFor(EPI.MainLoopVF); const auto &[ExpandedSCEVs, ReductionResumeValues] = LVP.executePlan( EPI.MainLoopVF, EPI.MainLoopUF, BestMainPlan, MainILV, DT, true); ++LoopsVectorized; // Second pass vectorizes the epilogue and adjusts the control flow // edges from the first pass. EPI.MainLoopVF = EPI.EpilogueVF; EPI.MainLoopUF = EPI.EpilogueUF; EpilogueVectorizerEpilogueLoop EpilogILV(L, PSE, LI, DT, TLI, TTI, AC, ORE, EPI, &LVL, &CM, BFI, PSI, Checks); VPlan &BestEpiPlan = LVP.getBestPlanFor(EPI.EpilogueVF); VPRegionBlock *VectorLoop = BestEpiPlan.getVectorLoopRegion(); VPBasicBlock *Header = VectorLoop->getEntryBasicBlock(); Header->setName("vec.epilog.vector.body"); // Re-use the trip count and steps expanded for the main loop, as // skeleton creation needs it as a value that dominates both the scalar // and vector epilogue loops // TODO: This is a workaround needed for epilogue vectorization and it // should be removed once induction resume value creation is done // directly in VPlan. EpilogILV.setTripCount(MainILV.getTripCount()); for (auto &R : make_early_inc_range(*BestEpiPlan.getPreheader())) { auto *ExpandR = cast(&R); auto *ExpandedVal = BestEpiPlan.getVPValueOrAddLiveIn( ExpandedSCEVs.find(ExpandR->getSCEV())->second); ExpandR->replaceAllUsesWith(ExpandedVal); ExpandR->eraseFromParent(); } // Ensure that the start values for any VPWidenIntOrFpInductionRecipe, // VPWidenPointerInductionRecipe and VPReductionPHIRecipes are updated // before vectorizing the epilogue loop. for (VPRecipeBase &R : Header->phis()) { if (isa(&R)) continue; Value *ResumeV = nullptr; // TODO: Move setting of resume values to prepareToExecute. if (auto *ReductionPhi = dyn_cast(&R)) { ResumeV = ReductionResumeValues .find(&ReductionPhi->getRecurrenceDescriptor()) ->second; } else { // Create induction resume values for both widened pointer and // integer/fp inductions and update the start value of the induction // recipes to use the resume value. PHINode *IndPhi = nullptr; const InductionDescriptor *ID; if (auto *Ind = dyn_cast(&R)) { IndPhi = cast(Ind->getUnderlyingValue()); ID = &Ind->getInductionDescriptor(); } else { auto *WidenInd = cast(&R); IndPhi = WidenInd->getPHINode(); ID = &WidenInd->getInductionDescriptor(); } ResumeV = MainILV.createInductionResumeValue( IndPhi, *ID, getExpandedStep(*ID, ExpandedSCEVs), {EPI.MainLoopIterationCountCheck}); } assert(ResumeV && "Must have a resume value"); VPValue *StartVal = BestEpiPlan.getVPValueOrAddLiveIn(ResumeV); cast(&R)->setStartValue(StartVal); } LVP.executePlan(EPI.EpilogueVF, EPI.EpilogueUF, BestEpiPlan, EpilogILV, DT, true, &ExpandedSCEVs); ++LoopsEpilogueVectorized; if (!MainILV.areSafetyChecksAdded()) DisableRuntimeUnroll = true; } else { InnerLoopVectorizer LB(L, PSE, LI, DT, TLI, TTI, AC, ORE, VF.Width, VF.MinProfitableTripCount, IC, &LVL, &CM, BFI, PSI, Checks); VPlan &BestPlan = LVP.getBestPlanFor(VF.Width); LVP.executePlan(VF.Width, IC, BestPlan, LB, DT, false); ++LoopsVectorized; // Add metadata to disable runtime unrolling a scalar loop when there // are no runtime checks about strides and memory. A scalar loop that is // rarely used is not worth unrolling. if (!LB.areSafetyChecksAdded()) DisableRuntimeUnroll = true; } // Report the vectorization decision. reportVectorization(ORE, L, VF, IC); } if (ORE->allowExtraAnalysis(LV_NAME)) checkMixedPrecision(L, ORE); } std::optional RemainderLoopID = makeFollowupLoopID(OrigLoopID, {LLVMLoopVectorizeFollowupAll, LLVMLoopVectorizeFollowupEpilogue}); if (RemainderLoopID) { L->setLoopID(*RemainderLoopID); } else { if (DisableRuntimeUnroll) AddRuntimeUnrollDisableMetaData(L); // Mark the loop as already vectorized to avoid vectorizing again. Hints.setAlreadyVectorized(); } assert(!verifyFunction(*L->getHeader()->getParent(), &dbgs())); return true; } LoopVectorizeResult LoopVectorizePass::runImpl( Function &F, ScalarEvolution &SE_, LoopInfo &LI_, TargetTransformInfo &TTI_, DominatorTree &DT_, BlockFrequencyInfo *BFI_, TargetLibraryInfo *TLI_, DemandedBits &DB_, AssumptionCache &AC_, LoopAccessInfoManager &LAIs_, OptimizationRemarkEmitter &ORE_, ProfileSummaryInfo *PSI_) { SE = &SE_; LI = &LI_; TTI = &TTI_; DT = &DT_; BFI = BFI_; TLI = TLI_; AC = &AC_; LAIs = &LAIs_; DB = &DB_; ORE = &ORE_; PSI = PSI_; // Don't attempt if // 1. the target claims to have no vector registers, and // 2. interleaving won't help ILP. // // The second condition is necessary because, even if the target has no // vector registers, loop vectorization may still enable scalar // interleaving. if (!TTI->getNumberOfRegisters(TTI->getRegisterClassForType(true)) && TTI->getMaxInterleaveFactor(ElementCount::getFixed(1)) < 2) return LoopVectorizeResult(false, false); bool Changed = false, CFGChanged = false; // The vectorizer requires loops to be in simplified form. // Since simplification may add new inner loops, it has to run before the // legality and profitability checks. This means running the loop vectorizer // will simplify all loops, regardless of whether anything end up being // vectorized. for (const auto &L : *LI) Changed |= CFGChanged |= simplifyLoop(L, DT, LI, SE, AC, nullptr, false /* PreserveLCSSA */); // Build up a worklist of inner-loops to vectorize. This is necessary as // the act of vectorizing or partially unrolling a loop creates new loops // and can invalidate iterators across the loops. SmallVector Worklist; for (Loop *L : *LI) collectSupportedLoops(*L, LI, ORE, Worklist); LoopsAnalyzed += Worklist.size(); // Now walk the identified inner loops. while (!Worklist.empty()) { Loop *L = Worklist.pop_back_val(); // For the inner loops we actually process, form LCSSA to simplify the // transform. Changed |= formLCSSARecursively(*L, *DT, LI, SE); Changed |= CFGChanged |= processLoop(L); if (Changed) { LAIs->clear(); #ifndef NDEBUG if (VerifySCEV) SE->verify(); #endif } } // Process each loop nest in the function. return LoopVectorizeResult(Changed, CFGChanged); } PreservedAnalyses LoopVectorizePass::run(Function &F, FunctionAnalysisManager &AM) { auto &LI = AM.getResult(F); // There are no loops in the function. Return before computing other expensive // analyses. if (LI.empty()) return PreservedAnalyses::all(); auto &SE = AM.getResult(F); auto &TTI = AM.getResult(F); auto &DT = AM.getResult(F); auto &TLI = AM.getResult(F); auto &AC = AM.getResult(F); auto &DB = AM.getResult(F); auto &ORE = AM.getResult(F); LoopAccessInfoManager &LAIs = AM.getResult(F); auto &MAMProxy = AM.getResult(F); ProfileSummaryInfo *PSI = MAMProxy.getCachedResult(*F.getParent()); BlockFrequencyInfo *BFI = nullptr; if (PSI && PSI->hasProfileSummary()) BFI = &AM.getResult(F); LoopVectorizeResult Result = runImpl(F, SE, LI, TTI, DT, BFI, &TLI, DB, AC, LAIs, ORE, PSI); if (!Result.MadeAnyChange) return PreservedAnalyses::all(); PreservedAnalyses PA; if (isAssignmentTrackingEnabled(*F.getParent())) { for (auto &BB : F) RemoveRedundantDbgInstrs(&BB); } // We currently do not preserve loopinfo/dominator analyses with outer loop // vectorization. Until this is addressed, mark these analyses as preserved // only for non-VPlan-native path. // TODO: Preserve Loop and Dominator analyses for VPlan-native path. if (!EnableVPlanNativePath) { PA.preserve(); PA.preserve(); PA.preserve(); } if (Result.MadeCFGChange) { // Making CFG changes likely means a loop got vectorized. Indicate that // extra simplification passes should be run. // TODO: MadeCFGChanges is not a prefect proxy. Extra passes should only // be run if runtime checks have been added. AM.getResult(F); PA.preserve(); } else { PA.preserveSet(); } return PA; } void LoopVectorizePass::printPipeline( raw_ostream &OS, function_ref MapClassName2PassName) { static_cast *>(this)->printPipeline( OS, MapClassName2PassName); OS << '<'; OS << (InterleaveOnlyWhenForced ? "" : "no-") << "interleave-forced-only;"; OS << (VectorizeOnlyWhenForced ? "" : "no-") << "vectorize-forced-only;"; OS << '>'; }