//===- 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 "VPlanCFG.h" #include "VPlanHelpers.h" #include "VPlanPatternMatch.h" #include "VPlanTransforms.h" #include "VPlanUtils.h" #include "VPlanVerifier.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/SmallVector.h" #include "llvm/ADT/Statistic.h" #include "llvm/ADT/StringRef.h" #include "llvm/ADT/Twine.h" #include "llvm/ADT/TypeSwitch.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/ScalarEvolutionPatternMatch.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/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/Verifier.h" #include "llvm/Support/Casting.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/Debug.h" #include "llvm/Support/ErrorHandling.h" #include "llvm/Support/InstructionCost.h" #include "llvm/Support/MathExtras.h" #include "llvm/Support/NativeFormatting.h" #include "llvm/Support/raw_ostream.h" #include "llvm/Transforms/Utils/BasicBlockUtils.h" #include "llvm/Transforms/Utils/InjectTLIMappings.h" #include "llvm/Transforms/Utils/Local.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 using namespace llvm; using namespace SCEVPatternMatch; using namespace LoopVectorizationUtils; #define LV_NAME "loop-vectorize" #define DEBUG_TYPE LV_NAME #ifndef NDEBUG const char VerboseDebug[] = DEBUG_TYPE "-verbose"; #endif STATISTIC(LoopsVectorized, "Number of loops vectorized"); STATISTIC(LoopsAnalyzed, "Number of loops analyzed for vectorization"); STATISTIC(LoopsEpilogueVectorized, "Number of epilogues vectorized"); STATISTIC(LoopsEarlyExitVectorized, "Number of early exit loops vectorized"); STATISTIC(LoopsPartialAliasVectorized, "Number of partial aliasing loops 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(ElementCount::getFixed(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. Note: This allows all scalable VFs >= vscale x 1.")); static cl::opt EpilogueVectorizationMinVF( "epilogue-vectorization-minimum-VF", 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")); static cl::opt ForcePartialAliasingVectorization( "force-partial-aliasing-vectorization", cl::init(false), cl::Hidden, cl::desc("Replace pointer diff checks with alias masks.")); /// Option tail-folding-policy controls the tail-folding strategy and lists all /// available options. The vectorizer will attempt to fold the tail-loop into /// the vector loop (main/epilogue loops) and predicate the instructions /// accordingly. If tail-folding fails, there are different fallback strategies /// depending on these values: enum class TailFoldingPolicyTy { None = 0, PreferFoldTail, MustFoldTail }; static cl::opt TailFoldingPolicy( "tail-folding-policy", cl::init(TailFoldingPolicyTy::None), cl::Hidden, cl::desc("Tail-folding preferences over creating an epilogue loop."), cl::values( clEnumValN(TailFoldingPolicyTy::None, "dont-fold-tail", "Don't tail-fold loops."), clEnumValN(TailFoldingPolicyTy::PreferFoldTail, "prefer-fold-tail", "prefer tail-folding, otherwise create an epilogue when " "appropriate."), clEnumValN(TailFoldingPolicyTy::MustFoldTail, "must-fold-tail", "always tail-fold, don't attempt vectorization if " "tail-folding fails."))); static cl::opt EpilogueTailFoldingPolicy( "epilogue-tail-folding-policy", cl::Hidden, cl::desc( "Epilogue-tail-folding preferences over creating an epilogue loop."), cl::values( clEnumValN(TailFoldingPolicyTy::None, "dont-fold-tail", "Don't tail-fold loops."), clEnumValN(TailFoldingPolicyTy::PreferFoldTail, "prefer-fold-tail", "prefer tail-folding, otherwise create an epilogue when " "appropriate."))); 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::DataWithEVL, "data-with-evl", "Use predicated EVL instructions for tail folding. If EVL " "is unsupported, fallback to data-without-lane-mask."))); cl::opt llvm::EnableWideActiveLaneMask( "enable-wide-lane-mask", cl::init(false), cl::Hidden, cl::desc("Enable use of wide lane masks when used for control flow in " "tail-folded loops")); 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 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.")); cl::opt llvm::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 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")); /// The number of stores in a loop that are allowed to need predication. cl::opt NumberOfStoresToPredicate( "vectorize-num-stores-pred", cl::init(1), cl::Hidden, cl::desc("Max number of stores to be predicated behind an if.")); // TODO: Move size-based thresholds out of legality checking, make cost based // decisions instead of hard thresholds. static cl::opt VectorizeSCEVCheckThreshold( "vectorize-scev-check-threshold", cl::init(16), cl::Hidden, cl::desc("The maximum number of SCEV checks allowed.")); static cl::opt PragmaVectorizeSCEVCheckThreshold( "pragma-vectorize-scev-check-threshold", cl::init(128), cl::Hidden, cl::desc("The maximum number of SCEV checks allowed with a " "vectorize(enable) pragma")); 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 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 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.")); cl::opt llvm::EnableVPlanNativePath( "enable-vplan-native-path", cl::Hidden, cl::desc("Enable VPlan-native vectorization path with " "support for outer loop vectorization.")); cl::opt llvm::VerifyEachVPlan("vplan-verify-each", #ifdef EXPENSIVE_CHECKS cl::init(true), #else cl::init(false), #endif cl::Hidden, cl::desc("Verify VPlans after VPlan transforms.")); #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP) cl::opt llvm::VPlanPrintBeforeAll( "vplan-print-before-all", cl::init(false), cl::Hidden, cl::desc("Print VPlans before all VPlan transformations.")); cl::opt llvm::VPlanPrintAfterAll( "vplan-print-after-all", cl::init(false), cl::Hidden, cl::desc("Print VPlans after all VPlan transformations.")); cl::list llvm::VPlanPrintBeforePasses( "vplan-print-before", cl::Hidden, cl::desc("Print VPlans before specified VPlan transformations (regexp).")); cl::list llvm::VPlanPrintAfterPasses( "vplan-print-after", cl::Hidden, cl::desc("Print VPlans after specified VPlan transformations (regexp).")); cl::opt llvm::VPlanPrintVectorRegionScope( "vplan-print-vector-region-scope", cl::init(false), cl::Hidden, cl::desc("Limit VPlan printing to vector loop region in " "`-vplan-print-after*` if the plan has one.")); #endif // 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. cl::opt VPlanBuildOuterloopStressTest( "vplan-build-outerloop-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 ForceMaskedDivRem("force-widen-divrem-via-masked-intrinsic", cl::Hidden, cl::desc("Override cost based masked intrinsic widening " "for div/rem instructions")); static cl::opt EnableEarlyExitVectorization( "enable-early-exit-vectorization", cl::init(true), cl::Hidden, cl::desc( "Enable vectorization of early exit loops with uncountable exits.")); static cl::opt EnableEarlyExitVectorizationWithSideEffects( "enable-early-exit-vectorization-with-side-effects", cl::init(false), cl::Hidden, cl::desc("Enable vectorization of early exit loops with uncountable exits " "and side effects")); // Returns true if the epilogue VF has been set to a non-zero value other than // VF=1 (scalar). static bool hasForcedEpilogueVF() { return EpilogueVectorizationForceVF.isNonZero() && EpilogueVectorizationForceVF != ElementCount::getFixed(1); } // Likelyhood of bypassing the vectorized loop because there are zero trips left // after prolog. See `emitIterationCountCheck`. static constexpr uint32_t MinItersBypassWeights[] = {1, 127}; /// A version of ScalarEvolution::getSmallConstantTripCount that returns an /// ElementCount to include loops whose trip count is a function of vscale. static ElementCount getSmallConstantTripCount(ScalarEvolution *SE, const Loop *L) { if (unsigned ExpectedTC = SE->getSmallConstantTripCount(L)) return ElementCount::getFixed(ExpectedTC); const SCEV *BTC = SE->getBackedgeTakenCount(L); if (isa(BTC)) return ElementCount::getFixed(0); const SCEV *ExitCount = SE->getTripCountFromExitCount(BTC, BTC->getType(), L); if (isa(ExitCount)) return ElementCount::getScalable(1); const APInt *Scale; if (match(ExitCount, m_scev_Mul(m_scev_APInt(Scale), m_SCEVVScale()))) if (cast(ExitCount)->hasNoUnsignedWrap()) if (Scale->getActiveBits() <= 32) return ElementCount::getScalable(Scale->getZExtValue()); return ElementCount::getFixed(0); } /// Get the maximum trip count for \p L from the SCEV unsigned range, excluding /// zero from the range. Only valid when not folding the tail, as the minimum /// iteration count check guards against a zero trip count. Returns 0 if /// unknown. static unsigned getMaxTCFromNonZeroRange(PredicatedScalarEvolution &PSE, Loop *L) { const SCEV *BTC = PSE.getBackedgeTakenCount(); if (isa(BTC)) return 0; ScalarEvolution *SE = PSE.getSE(); const SCEV *TripCount = SE->getTripCountFromExitCount(BTC, BTC->getType(), L); ConstantRange TCRange = SE->getUnsignedRange(TripCount); APInt MaxTCFromRange = TCRange.getUnsignedMax(); if (!MaxTCFromRange.isZero() && MaxTCFromRange.getActiveBits() <= 32) return MaxTCFromRange.getZExtValue(); return 0; } /// Returns "best known" trip count, which is either a valid positive trip count /// or std::nullopt when an estimate cannot be made (including when the trip /// count would overflow), 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 known, if \p CanUseConstantMax, and /// if \p ComputeUpperBoundOnly is false. /// 4) Returns the maximum trip count from the SCEV range excluding zero, /// if \p CanUseConstantMax and \p CanExcludeZeroTrips. /// 5) Returns std::nullopt if all of the above failed. static std::optional getSmallBestKnownTC( PredicatedScalarEvolution &PSE, Loop *L, bool CanUseConstantMax = true, bool CanExcludeZeroTrips = false, bool ComputeUpperBoundOnly = false) { // Check if exact trip count is known. if (auto ExpectedTC = getSmallConstantTripCount(PSE.getSE(), L)) return ExpectedTC; // Check if there is an expected trip count available from profile data. if (LoopVectorizeWithBlockFrequency && !ComputeUpperBoundOnly) if (auto EstimatedTC = getLoopEstimatedTripCount(L)) return ElementCount::getFixed(*EstimatedTC); if (!CanUseConstantMax) return std::nullopt; // Check if upper bound estimate is known. if (unsigned ExpectedTC = PSE.getSmallConstantMaxTripCount()) return ElementCount::getFixed(ExpectedTC); // Get the maximum trip count from the SCEV range excluding zero. This is // only safe when not folding the tail, as the minimum iteration count check // prevents entering the vector loop with a zero trip count. if (CanUseConstantMax && CanExcludeZeroTrips) if (unsigned RefinedTC = getMaxTCFromNonZeroRange(PSE, L)) return ElementCount::getFixed(RefinedTC); return std::nullopt; } 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 TargetTransformInfo *TTI, AssumptionCache *AC, ElementCount VecWidth, unsigned UnrollFactor, LoopVectorizationCostModel *CM, GeneratedRTChecks &RTChecks, VPlan &Plan) : OrigLoop(OrigLoop), PSE(PSE), LI(LI), DT(DT), TTI(TTI), AC(AC), VF(VecWidth), UF(UnrollFactor), Builder(PSE.getSE()->getContext()), Cost(CM), RTChecks(RTChecks), Plan(Plan), VectorPHVPBB(cast( Plan.getVectorLoopRegion()->getSinglePredecessor())) {} virtual ~InnerLoopVectorizer() = default; /// Creates a basic block for the scalar preheader. Both /// EpilogueVectorizerMainLoop and EpilogueVectorizerEpilogueLoop overwrite /// the method to create additional blocks and checks needed for epilogue /// vectorization. virtual BasicBlock *createVectorizedLoopSkeleton(); /// Fix the vectorized code, taking care of header phi's, and more. void fixVectorizedLoop(VPTransformState &State); protected: friend class LoopVectorizationPlanner; /// Create and return a new IR basic block for the scalar preheader whose name /// is prefixed with \p Prefix. BasicBlock *createScalarPreheader(StringRef Prefix); /// 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 Transform Info. const TargetTransformInfo *TTI; /// Assumption Cache. AssumptionCache *AC; /// The vectorization SIMD factor to use. Each vector will have this many /// vector elements. ElementCount VF; /// 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 profitablity analysis. LoopVectorizationCostModel *Cost; /// Structure to hold information about generated runtime checks, responsible /// for cleaning the checks, if vectorization turns out unprofitable. GeneratedRTChecks &RTChecks; VPlan &Plan; /// The vector preheader block of \p Plan, used as target for check blocks /// introduced during skeleton creation. VPBasicBlock *VectorPHVPBB; }; /// 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; Value *VectorTripCount = nullptr; VPlan &EpiloguePlan; EpilogueLoopVectorizationInfo(ElementCount MVF, unsigned MUF, ElementCount EVF, unsigned EUF, VPlan &EpiloguePlan) : MainLoopVF(MVF), MainLoopUF(MUF), EpilogueVF(EVF), EpilogueUF(EUF), EpiloguePlan(EpiloguePlan) { 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 TargetTransformInfo *TTI, AssumptionCache *AC, EpilogueLoopVectorizationInfo &EPI, LoopVectorizationCostModel *CM, GeneratedRTChecks &Checks, VPlan &Plan, ElementCount VecWidth, unsigned UnrollFactor) : InnerLoopVectorizer(OrigLoop, PSE, LI, DT, TTI, AC, VecWidth, UnrollFactor, CM, Checks, Plan), EPI(EPI) {} /// 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 TargetTransformInfo *TTI, AssumptionCache *AC, EpilogueLoopVectorizationInfo &EPI, LoopVectorizationCostModel *CM, GeneratedRTChecks &Check, VPlan &Plan) : InnerLoopAndEpilogueVectorizer(OrigLoop, PSE, LI, DT, TTI, AC, EPI, CM, Check, Plan, EPI.MainLoopVF, EPI.MainLoopUF) {} protected: 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 TargetTransformInfo *TTI, AssumptionCache *AC, EpilogueLoopVectorizationInfo &EPI, LoopVectorizationCostModel *CM, GeneratedRTChecks &Checks, VPlan &Plan) : InnerLoopAndEpilogueVectorizer(OrigLoop, PSE, LI, DT, TTI, AC, EPI, CM, Checks, Plan, EPI.EpilogueVF, EPI.EpilogueUF) {} /// Implements the interface for creating a vectorized skeleton using the /// *epilogue loop* strategy (i.e., the second pass of VPlan execution). BasicBlock *createVectorizedLoopSkeleton() final; protected: void printDebugTracesAtStart() override; void printDebugTracesAtEnd() override; }; } // end namespace llvm /// Look for a meaningful debug location on the instruction or its operands. static DebugLoc getDebugLocFromInstOrOperands(Instruction *I) { if (!I) return DebugLoc::getUnknown(); 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(); } namespace llvm { /// Return the runtime value for VF. Value *getRuntimeVF(IRBuilderBase &B, Type *Ty, ElementCount VF) { return B.CreateElementCount(Ty, VF); } } // end namespace llvm namespace llvm { // Loop vectorization cost-model hints how the epilogue/tail loop should be // lowered. enum EpilogueLowering { // The default: allowing epilogues. CM_EpilogueAllowed, // Vectorization with OptForSize: don't allow epilogues. CM_EpilogueNotAllowedOptSize, // 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_EpilogueNotAllowedLowTripLoop, // Loop hint indicating an epilogue is undesired, apply tail folding. CM_EpilogueNotNeededFoldTail, // Directive indicating we must either fold the epilogue/tail or not vectorize CM_EpilogueNotAllowedFoldTail }; enum class AliasMaskingStatus { NotDecided, Disabled, Enabled }; /// 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 { friend class LoopVectorizationPlanner; public: LoopVectorizationCostModel(EpilogueLowering SEL, Loop *L, PredicatedScalarEvolution &PSE, LoopInfo *LI, LoopVectorizationLegality *Legal, const TargetTransformInfo &TTI, const TargetLibraryInfo *TLI, AssumptionCache *AC, OptimizationRemarkEmitter *ORE, std::function GetBFI, const Function *F, const LoopVectorizeHints *Hints, InterleavedAccessInfo &IAI, VFSelectionContext &Config) : Config(Config), EpilogueLoweringStatus(SEL), TheLoop(L), PSE(PSE), LI(LI), Legal(Legal), TTI(TTI), TLI(TLI), AC(AC), ORE(ORE), GetBFI(GetBFI), 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); /// 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); /// Collect values we want to ignore in the cost model. void collectValuesToIgnore(); /// \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."); assert( TheLoop->isInnermost() && "cost-model should not be used for outer loops (in VPlan-native path)"); 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 { assert( TheLoop->isInnermost() && "cost-model should not be used for outer loops (in VPlan-native path)"); // If VF is scalar, then all instructions are trivially uniform. if (VF.isScalar()) return true; // Pseudo probes must be duplicated per vector lane so that the // profiled loop trip count is not undercounted. if (isa(I)) 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 { assert( TheLoop->isInnermost() && "cost-model should not be used for outer loops (in VPlan-native path)"); if (VF.isScalar()) return true; 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 { const auto &MinBWs = Config.getMinimalBitwidths(); // Truncs must truncate at most to their destination type. if (isa_and_nonnull(I) && MinBWs.contains(I) && I->getType()->getScalarSizeInBits() < MinBWs.lookup(I)) return false; 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, /// A widening decision that has been invalidated after replacing the /// corresponding recipe during VPlan transforms. /// TODO: Remove once the legacy exit cost computation is retired. CM_InvalidatedDecision }; /// 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[{I, VF}] = {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. /// When interleaving, the cost will only be assigned one instruction, the /// insert position. For other cases, add the appropriate fraction of the /// total cost to each instruction. This ensures accurate costs are used, /// even if the insert position instruction is not used. InstructionCost InsertPosCost = Cost; InstructionCost OtherMemberCost = 0; if (W != CM_Interleave) OtherMemberCost = InsertPosCost = Cost / Grp->getNumMembers(); ; for (auto *I : Grp->members()) { if (Grp->getInsertPos() == I) WideningDecisions[{I, VF}] = {W, InsertPosCost}; else WideningDecisions[{I, VF}] = {W, OtherMemberCost}; } } /// 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"); assert( TheLoop->isInnermost() && "cost-model should not be used for outer loops (in VPlan-native path)"); std::pair InstOnVF(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(I, VF); assert(WideningDecisions.contains(InstOnVF) && "The cost is not calculated"); return WideningDecisions[InstOnVF].second; } /// 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(Trunc->getSrcTy(), VF); Type *DestTy = toVectorTy(Trunc->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 values that will not be widened, including Uniforms, Scalars, and /// Instructions to Scalarize 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 collectNonVectorizedAndSetWideningDecisions(ElementCount VF) { // Do the analysis once. if (VF.isScalar() || Uniforms.contains(VF)) return; setCostBasedWideningDecision(VF); collectLoopUniforms(VF); collectLoopScalars(VF); collectInstsToScalarize(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 MaskedCost) const { switch (ForceMaskedDivRem) { case cl::boolOrDefault::BOU_UNSET: return ScalarCost < MaskedCost; case cl::boolOrDefault::BOU_TRUE: return false; case cl::boolOrDefault::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); /// Wrapper function for LoopVectorizationLegality::isMaskRequired, /// that passes the Instruction \p I and if we fold tail. bool isMaskRequired(Instruction *I) 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; /// A helper function that returns how much we should divide the cost of a /// predicated block by. Typically this is the reciprocal of the block /// probability, i.e. if we return X we are assuming the predicated block will /// execute once for every X iterations of the loop header so the block should /// only contribute 1/X of its cost to the total cost calculation, but when /// optimizing for code size it will just be 1 as code size costs don't depend /// on execution probabilities. /// /// Note that if a block wasn't originally predicated but was predicated due /// to tail folding, the divisor will still be 1 because it will execute for /// every iteration of the loop header. inline uint64_t getPredBlockCostDivisor(TargetTransformInfo::TargetCostKind CostKind, const BasicBlock *BB); /// Returns true if an artificially high cost for emulated masked memrefs /// should be used. bool useEmulatedMaskMemRefHack(Instruction *I, ElementCount VF); /// 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 masked intrinsic strategy. std::pair getDivRemSpeculationCost(Instruction *I, ElementCount VF); /// If \p I is a memory instruction with a consecutive pointer that can be /// widened, returns the widening kind (CM_Widen or CM_Widen_Reverse) and /// std::nullopt otherwise. std::optional 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) const; /// Check if \p Instr belongs to any interleaved access group. bool isAccessInterleaved(Instruction *Instr) const { return InterleaveInfo.isInterleaved(Instr); } /// Get the interleaved access group that \p Instr belongs to. const InterleaveGroup * getInterleavedAccessGroup(Instruction *Instr) const { 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 (!isEpilogueAllowed()) { LLVM_DEBUG(dbgs() << "LV: Loop does not require scalar epilogue\n"); return false; } // If we might exit from anywhere but the latch and early exit vectorization // is disabled, we must run the exiting iteration in scalar form. if (TheLoop->getExitingBlock() != TheLoop->getLoopLatch() && !(EnableEarlyExitVectorization && Legal->hasUncountableEarlyExit())) { LLVM_DEBUG(dbgs() << "LV: Loop requires scalar epilogue: not exiting " "from latch block\n"); return true; } if (IsVectorizing && InterleaveInfo.requiresScalarEpilogue()) { LLVM_DEBUG(dbgs() << "LV: Loop requires scalar epilogue: " "interleaved group requires scalar epilogue\n"); return true; } LLVM_DEBUG(dbgs() << "LV: Loop does not require scalar epilogue\n"); return false; } /// Returns true if an epilogue is allowed (e.g., not prevented by /// optsize or a loop hint annotation). bool isEpilogueAllowed() const { return EpilogueLoweringStatus == CM_EpilogueAllowed; } /// Returns true if tail-folding is preferred over an epilogue. bool preferTailFoldedLoop() const { return EpilogueLoweringStatus == CM_EpilogueNotNeededFoldTail || EpilogueLoweringStatus == CM_EpilogueNotAllowedFoldTail; } /// Returns the TailFoldingStyle that is best for the current loop. TailFoldingStyle getTailFoldingStyle() const { return ChosenTailFoldingStyle; } /// Selects and saves TailFoldingStyle. /// \param IsScalableVF true if scalable vector factors enabled. /// \param UserIC User specific interleave count. void setTailFoldingStyle(bool IsScalableVF, unsigned UserIC) { assert(ChosenTailFoldingStyle == TailFoldingStyle::None && "Tail folding must not be selected yet."); if (!Legal->canFoldTailByMasking()) { ChosenTailFoldingStyle = TailFoldingStyle::None; return; } // Default to TTI preference, but allow command line override. ChosenTailFoldingStyle = TTI.getPreferredTailFoldingStyle(); if (ForceTailFoldingStyle.getNumOccurrences()) ChosenTailFoldingStyle = ForceTailFoldingStyle.getValue(); if (ChosenTailFoldingStyle != TailFoldingStyle::DataWithEVL) return; // Override EVL styles if needed. // FIXME: Investigate opportunity for fixed vector factor. bool EVLIsLegal = UserIC <= 1 && IsScalableVF && TTI.hasActiveVectorLength() && !EnableVPlanNativePath; if (EVLIsLegal) return; // If for some reason EVL mode is unsupported, fallback to an epilogue // if it's allowed, or DataWithoutLaneMask otherwise. if (EpilogueLoweringStatus == CM_EpilogueAllowed || EpilogueLoweringStatus == CM_EpilogueNotNeededFoldTail) ChosenTailFoldingStyle = TailFoldingStyle::None; else ChosenTailFoldingStyle = TailFoldingStyle::DataWithoutLaneMask; LLVM_DEBUG( dbgs() << "LV: Preference for VP intrinsics indicated. Will " "not try to generate VP Intrinsics " << (UserIC > 1 ? "since interleave count specified is greater than 1.\n" : "due to non-interleaving reasons.\n")); } /// Returns true if all loop blocks should be masked to fold tail loop. bool foldTailByMasking() const { return getTailFoldingStyle() != TailFoldingStyle::None; } void tryToEnablePartialAliasMasking() { assert(foldTailByMasking() && "Expected tail folding to be enabled!"); assert(!foldTailWithEVL() && "Did not expect to enable alias masking with EVL!"); assert(PartialAliasMaskingStatus == AliasMaskingStatus::NotDecided); // Assume we fail to enable alias masking (in case we early exit). PartialAliasMaskingStatus = AliasMaskingStatus::Disabled; // Note: FixedOrderRecurrences are not supported yet as we cannot handle // the required `splice.right` with the alias-mask. if (!ForcePartialAliasingVectorization || !Legal->getFixedOrderRecurrences().empty()) return; const RuntimePointerChecking *Checks = Legal->getRuntimePointerChecking(); if (!Checks) return; auto DiffChecks = Checks->getDiffChecks(); if (!DiffChecks || DiffChecks->empty()) return; [[maybe_unused]] auto HasPointerArgs = [](CallBase *CB) { return any_of(CB->args(), [](Value const *Arg) { return Arg->getType()->isPointerTy(); }); }; for (BasicBlock *BB : TheLoop->blocks()) { for (Instruction &I : *BB) { if (!isa(I)) { [[maybe_unused]] auto *Call = dyn_cast(&I); assert( (!I.mayReadOrWriteMemory() || (Call && !HasPointerArgs(Call))) && "Skipped unexpected memory access"); continue; } Type *ScalarTy = getLoadStoreType(&I); Value *Ptr = getLoadStorePointerOperand(&I); // Currently, we can't handle alias masking in reverse. Reversing the // alias mask is not correct (or necessary). When combined with // tail-folding the active lane mask should only be reversed where the // alias-mask is true. if (Legal->isConsecutivePtr(ScalarTy, Ptr) == -1) return; } } PartialAliasMaskingStatus = AliasMaskingStatus::Enabled; } /// Returns true if all loop blocks should have partial aliases masked. bool maskPartialAliasing() const { return PartialAliasMaskingStatus == AliasMaskingStatus::Enabled; } /// Returns true if the use of wide lane masks is requested and the loop is /// using tail-folding with a lane mask for control flow. bool useWideActiveLaneMask() const { if (!EnableWideActiveLaneMask) return false; return getTailFoldingStyle() == TailFoldingStyle::DataAndControlFlow; } /// 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 VP intrinsics with explicit vector length support should /// be generated in the tail folded loop. bool foldTailWithEVL() const { return getTailFoldingStyle() == TailFoldingStyle::DataWithEVL; } /// Returns true if the predicated reduction select should be used to set the /// incoming value for the reduction phi. bool usePredicatedReductionSelect(RecurKind RecurrenceKind) const { // Force to use predicated reduction select since the EVL of the // second-to-last iteration might not be VF*UF. if (foldTailWithEVL()) return true; // Force a predicated select with alias-masking to avoid propagating poison // values to the header phi for lanes outside the alias-mask. if (maskPartialAliasing()) return true; // Note: For FindLast recurrences we prefer a predicated select to simplify // matching in handleFindLastReductions(), rather than handle multiple // cases. if (RecurrenceDescriptor::isFindLastRecurrenceKind(RecurrenceKind)) return true; return PreferPredicatedReductionSelect || TTI.preferPredicatedReductionSelect(); } /// 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(); Uniforms.clear(); Scalars.clear(); } /// 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. InstructionCost expectedCost(ElementCount VF); /// Returns true if epilogue vectorization is considered profitable, and /// false otherwise. /// \p VF is the vectorization factor chosen for the original loop. /// \p Multiplier is an aditional scaling factor applied to VF before /// comparing to EpilogueVectorizationMinVF. bool isEpilogueVectorizationProfitable(const ElementCount VF, const unsigned IC) const; /// Returns the execution time cost of an instruction for a given vector /// width. Vector width of one means scalar. InstructionCost getInstructionCost(Instruction *I, ElementCount VF); /// 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) const; /// Returns true if \p Op should be considered invariant and if it is /// trivially hoistable. bool shouldConsiderInvariant(Value *Op); /// Returns true if \p I has been forced to be scalarized at \p VF. bool isForcedScalar(Instruction *I, ElementCount VF) const { auto FS = ForcedScalars.find(VF); return FS != ForcedScalars.end() && FS->second.contains(I); } private: unsigned NumPredStores = 0; /// VF selection state independent of cost-modeling decisions. VFSelectionContext &Config; /// Wrapper around LoopVectorizationLegality::isUniform() that takes into /// account if alias-masking is enabled. We consider the VF to be unknown when /// alias masking. bool isUniform(Value *V, ElementCount VF) const { // With alias-masking our runtime VF is [2, VF] (and not necessarily a // power-of-two). Something that is uniform for VF may not be for the full // range. assert(PartialAliasMaskingStatus != AliasMaskingStatus::NotDecided && "alias-mask status must be decided already"); return Legal->isUniform(V, PartialAliasMaskingStatus == AliasMaskingStatus::Disabled ? std::optional(VF) : std::nullopt); } /// Wrapper around LoopVectorizationLegality::isUniformMemOp() that takes into /// account if alias-masking is enabled. We consider the VF to be unknown when /// alias masking. bool isUniformMemOp(Instruction &I, ElementCount VF) const { assert(PartialAliasMaskingStatus != AliasMaskingStatus::NotDecided && "alias-mask status must be decided already"); return Legal->isUniformMemOp(I, PartialAliasMaskingStatus == AliasMaskingStatus::Disabled ? std::optional(VF) : std::nullopt); } /// 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, InstWidening Kind); /// 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) const; /// A type representing the costs for instructions if they were to be /// scalarized rather than vectorized. The entries are Instruction-Cost /// pairs. using ScalarCostsTy = MapVector; /// 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. EpilogueLowering EpilogueLoweringStatus = CM_EpilogueAllowed; /// Control finally chosen tail folding style. TailFoldingStyle ChosenTailFoldingStyle = TailFoldingStyle::None; /// If partial alias masking is enabled/disabled or not decided. AliasMaskingStatus PartialAliasMaskingStatus = AliasMaskingStatus::NotDecided; /// 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. MapVector 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; /// 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; /// 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) || getWideningDecision(I, VF) == CM_Scalarize) 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 { SmallPtrSet UniqueOperands; SmallVector Res; for (Value *Op : Ops) { if (isa(Op) || !UniqueOperands.insert(Op).second || !needsExtract(Op, VF)) continue; Res.push_back(Op); } return Res; } 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; /// Assumption cache. AssumptionCache *AC; /// Interface to emit optimization remarks. OptimizationRemarkEmitter *ORE; /// A function to lazily fetch BlockFrequencyInfo. This avoids computing it /// unless necessary, e.g. when the loop isn't legal to vectorize or when /// there is no predication. std::function GetBFI; /// The BlockFrequencyInfo returned from GetBFI. BlockFrequencyInfo *BFI = nullptr; /// Returns the BlockFrequencyInfo for the function if cached, otherwise /// fetches it via GetBFI. Avoids an indirect call to the std::function. BlockFrequencyInfo &getBFI() { if (!BFI) BFI = &GetBFI(); return *BFI; } 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; }; } // 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 no SCEV checks have been generated. 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 no memory runtime checks have been generated. Value *MemRuntimeCheckCond = nullptr; DominatorTree *DT; LoopInfo *LI; TargetTransformInfo *TTI; SCEVExpander SCEVExp; SCEVExpander MemCheckExp; bool CostTooHigh = false; Loop *OuterLoop = nullptr; PredicatedScalarEvolution &PSE; /// The kind of cost that we are calculating TTI::TargetCostKind CostKind; /// True if the loop is alias-masked (which allows us to omit diff checks). bool LoopUsesPartialAliasMasking = false; public: GeneratedRTChecks(PredicatedScalarEvolution &PSE, DominatorTree *DT, LoopInfo *LI, TargetTransformInfo *TTI, TTI::TargetCostKind CostKind, bool LoopUsesPartialAliasMasking) : DT(DT), LI(LI), TTI(TTI), SCEVExp(*PSE.getSE(), "scev.check", /*PreserveLCSSA=*/false), MemCheckExp(*PSE.getSE(), "scev.check", /*PreserveLCSSA=*/false), PSE(PSE), CostKind(CostKind), LoopUsesPartialAliasMasking(LoopUsesPartialAliasMasking) {} /// 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 are 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, OptimizationRemarkEmitter &ORE) { // 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) { // Mark runtime checks as never succeeding when they exceed the threshold. MemRuntimeCheckCond = ConstantInt::getTrue(L->getHeader()->getContext()); SCEVCheckCond = ConstantInt::getTrue(L->getHeader()->getContext()); ORE.emit([&]() { return OptimizationRemarkAnalysisAliasing( DEBUG_TYPE, "TooManyMemoryRuntimeChecks", L->getStartLoc(), L->getHeader()) << "loop not vectorized: too many memory checks needed"; }); LLVM_DEBUG(dbgs() << "LV: Too many memory checks needed.\n"); 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()); if (isa(SCEVCheckCond)) { // Clean up directly after expanding the predicate to a constant, to // avoid further expansions re-using anything left over from SCEVExp. SCEVExpanderCleaner SCEVCleaner(SCEVExp); SCEVCleaner.cleanup(); } } const auto &RtPtrChecking = *LAI.getRuntimePointerChecking(); // TODO: We need to estimate the cost of alias-masking in // GeneratedRTChecks::getCost(). We can't check the MemCheckBlock as the // alias-mask is generated later in VPlan. if (RtPtrChecking.Need && !LoopUsesPartialAliasMasking) { 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"); } SCEVExp.eraseDeadInstructions(SCEVCheckCond); 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()->getIterator()); auto *UI = new UnreachableInst(Preheader->getContext(), SCEVCheckBlock); UI->setDebugLoc(DebugLoc::getTemporary()); Preheader->getTerminator()->eraseFromParent(); } if (MemCheckBlock) { MemCheckBlock->getTerminator()->moveBefore( Preheader->getTerminator()->getIterator()); auto *UI = new UnreachableInst(Preheader->getContext(), MemCheckBlock); UI->setDebugLoc(DebugLoc::getTemporary()); 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, CostKind); 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, CostKind); 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; // Get the best known TC estimate. if (auto EstimatedTC = getSmallBestKnownTC( PSE, OuterLoop, /* CanUseConstantMax = */ false)) if (EstimatedTC->isFixed()) BestTripCount = EstimatedTC->getFixedValue(); InstructionCost NewMemCheckCost = MemCheckCost / BestTripCount; // Let's ensure the cost is always at least 1. NewMemCheckCost = std::max(NewMemCheckCost.getValue(), (InstructionCost::CostType)1); if (BestTripCount > 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); bool SCEVChecksUsed = !SCEVCheckBlock || !pred_empty(SCEVCheckBlock); bool MemChecksUsed = !MemCheckBlock || !pred_empty(MemCheckBlock); if (SCEVChecksUsed) SCEVCleaner.markResultUsed(); if (MemChecksUsed) { MemCheckCleaner.markResultUsed(); } else { 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 (!SCEVChecksUsed) SCEVCheckBlock->eraseFromParent(); if (!MemChecksUsed) MemCheckBlock->eraseFromParent(); } /// Retrieves the SCEVCheckCond and SCEVCheckBlock that were generated as IR /// outside VPlan. std::pair getSCEVChecks() const { using namespace llvm::PatternMatch; if (!SCEVCheckCond || match(SCEVCheckCond, m_ZeroInt())) return {nullptr, nullptr}; return {SCEVCheckCond, SCEVCheckBlock}; } /// Retrieves the MemCheckCond and MemCheckBlock that were generated as IR /// outside VPlan. std::pair getMemRuntimeChecks() const { using namespace llvm::PatternMatch; if (MemRuntimeCheckCond && match(MemRuntimeCheckCond, m_ZeroInt())) return {nullptr, nullptr}; return {MemRuntimeCheckCond, MemCheckBlock}; } /// Return true if any runtime checks have been added bool hasChecks() const { return getSCEVChecks().first || getMemRuntimeChecks().first; } }; } // namespace static bool useActiveLaneMask(TailFoldingStyle Style) { return Style == TailFoldingStyle::Data || Style == TailFoldingStyle::DataAndControlFlow; } static bool useActiveLaneMaskForControlFlow(TailFoldingStyle Style) { return Style == TailFoldingStyle::DataAndControlFlow; } // 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() || VPlanBuildOuterloopStressTest || (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. //===----------------------------------------------------------------------===// /// 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 : std::max(Cost->TTI.getMaxInterleaveFactor(VF, false), Cost->TTI.getMaxInterleaveFactor(VF, true)); IntegerType *IdxTy = Cost->Legal->getWidestInductionType(); APInt MaxUIntTripCount = 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 (std::optional TC = getSmallBestKnownTC( Cost->PSE, Cost->TheLoop, /*CanUseConstantMax=*/true, /*CanExcludeZeroTrips=*/false, /*ComputeUpperBoundOnly=*/true)) { unsigned MaxVF = VF.getKnownMinValue(); unsigned MaxTC = TC->getKnownMinValue(); if (VF.isScalable() || TC->isScalable()) { std::optional MaxVScale = getMaxVScale(*Cost->TheFunction, Cost->TTI); if (!MaxVScale) return false; if (VF.isScalable()) MaxVF *= *MaxVScale; if (TC->isScalable()) { bool Overflow; MaxTC = SaturatingMultiply(MaxTC, *MaxVScale, &Overflow); if (Overflow) return false; } } return (MaxUIntTripCount - MaxTC).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(); } /// Replace \p VPBB with a VPIRBasicBlock wrapping \p IRBB. All recipes from \p /// VPBB are moved to the end of the newly created VPIRBasicBlock. All /// predecessors and successors of VPBB, if any, are rewired to the new /// VPIRBasicBlock. If \p VPBB may be unreachable, \p Plan must be passed. static VPIRBasicBlock *replaceVPBBWithIRVPBB(VPBasicBlock *VPBB, BasicBlock *IRBB, VPlan *Plan = nullptr) { if (!Plan) Plan = VPBB->getPlan(); VPIRBasicBlock *IRVPBB = Plan->createVPIRBasicBlock(IRBB); auto IP = IRVPBB->begin(); for (auto &R : make_early_inc_range(VPBB->phis())) R.moveBefore(*IRVPBB, IP); for (auto &R : make_early_inc_range(make_range(VPBB->getFirstNonPhi(), VPBB->end()))) R.moveBefore(*IRVPBB, IRVPBB->end()); VPBlockUtils::reassociateBlocks(VPBB, IRVPBB); // VPBB is now dead and will be cleaned up when the plan gets destroyed. return IRVPBB; } BasicBlock *InnerLoopVectorizer::createScalarPreheader(StringRef Prefix) { BasicBlock *VectorPH = OrigLoop->getLoopPreheader(); assert(VectorPH && "Invalid loop structure"); // NOTE: The Plan's scalar preheader VPBB isn't replaced with a VPIRBasicBlock // wrapping the newly created scalar preheader here at the moment, because the // Plan's scalar preheader may be unreachable at this point. Instead it is // replaced in executePlan. return SplitBlock(VectorPH, VectorPH->getTerminator(), DT, LI, nullptr, Twine(Prefix) + "scalar.ph"); } /// Knowing that loop \p L executes a single vector iteration, add instructions /// that will get simplified and thus should not have any cost to \p /// InstsToIgnore. static void addFullyUnrolledInstructionsToIgnore( Loop *L, const LoopVectorizationLegality::InductionList &IL, SmallPtrSetImpl &InstsToIgnore) { auto *Cmp = L->getLatchCmpInst(); if (Cmp) InstsToIgnore.insert(Cmp); for (const auto &KV : IL) { // Extract the key by hand so that it can be used in the lambda below. Note // that captured structured bindings are a C++20 extension. const PHINode *IV = KV.first; // Get next iteration value of the induction variable. Instruction *IVInst = cast(IV->getIncomingValueForBlock(L->getLoopLatch())); if (all_of(IVInst->users(), [&](const User *U) { return U == IV || U == Cmp; })) InstsToIgnore.insert(IVInst); } } BasicBlock *InnerLoopVectorizer::createVectorizedLoopSkeleton() { // Create a new IR basic block for the scalar preheader. BasicBlock *ScalarPH = createScalarPreheader(""); return ScalarPH->getSinglePredecessor(); } namespace { struct CSEDenseMapInfo { static bool canHandle(const Instruction *I) { return isa(I) || isa(I) || isa(I) || isa(I); } static unsigned getHashValue(const Instruction *I) { assert(canHandle(I) && "Unknown instruction!"); return hash_combine(I->getOpcode(), hash_combine_range(I->operand_values())); } static bool isEqual(const Instruction *LHS, const Instruction *RHS) { return LHS->isIdenticalTo(RHS); } }; } // end anonymous namespace /// FIXME: This legacy common-subexpression-elimination routine is scheduled for /// removal, in favor of the VPlan-based one. static void legacyCSE(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; } } /// This function attempts to return a value that represents the ElementCount /// at runtime. For fixed-width VFs we know this precisely at compile /// time, but for scalable VFs we calculate it based on an estimate of the /// vscale value. static unsigned estimateElementCount(ElementCount VF, std::optional VScale) { unsigned EstimatedVF = VF.getKnownMinValue(); if (VF.isScalable()) if (VScale) EstimatedVF *= *VScale; assert(EstimatedVF >= 1 && "Estimated VF shouldn't be less than 1"); return EstimatedVF; } /// Returns the vector library variant function of \p CI usable at \p VF, /// respecting \p MaskRequired, or nullptr if none is found: a mapping with /// matching VF, masked if required, whose vector function is declared in the /// module. static Function *getVectorLibraryVariantFor(const CallInst &CI, ElementCount VF, bool MaskRequired, const TargetLibraryInfo *TLI) { if (!TLI || CI.isNoBuiltin()) return nullptr; for (const VFInfo &Info : VFDatabase::getMappings(CI)) if (Info.Shape.VF == VF && (!MaskRequired || Info.isMasked())) if (Function *F = CI.getModule()->getFunction(Info.VectorName)) return F; return nullptr; } /// Returns true iff \p CI has a library vector variant usable at \p VF. static bool hasVectorLibraryVariantFor(const CallInst &CI, ElementCount VF, bool MaskRequired, const TargetLibraryInfo *TLI) { return getVectorLibraryVariantFor(CI, VF, MaskRequired, TLI) != nullptr; } InstructionCost LoopVectorizationCostModel::getVectorCallCost(CallInst *CI, ElementCount VF) const { Type *RetTy = CI->getType(); SmallVector Tys; for (auto &ArgOp : CI->args()) Tys.push_back(ArgOp->getType()); InstructionCost ScalarCallCost = TTI.getCallInstrCost( CI->getCalledFunction(), RetTy, Tys, Config.CostKind); // Cost of the scalar call (scalar VF) or its scalarization (vector VF). The // scalarization cost is only meaningful for fixed VFs. InstructionCost Cost = VF.isScalable() ? InstructionCost::getInvalid() : ScalarCallCost * VF.getKnownMinValue() + getScalarizationOverhead(CI, VF); // The call may be vectorized at this VF, via a vector intrinsic or a vector // library variant. if (getVectorIntrinsicIDForCall(CI, TLI)) Cost = std::min(Cost, getVectorIntrinsicCost(CI, VF)); if (Function *Variant = getVectorLibraryVariantFor(*CI, VF, isMaskRequired(CI), TLI)) Cost = std::min(Cost, TTI.getCallInstrCost( /*F=*/nullptr, Variant->getReturnType(), Variant->getFunctionType()->params(), Config.CostKind)); return Cost; } static Type *maybeVectorizeType(Type *Ty, ElementCount VF) { if (VF.isScalar() || !canVectorizeTy(Ty)) return Ty; return toVectorizedTy(Ty, 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), InstructionCost::getInvalid()); return TTI.getIntrinsicInstrCost(CostAttrs, Config.CostKind); } void InnerLoopVectorizer::fixVectorizedLoop(VPTransformState &State) { // Don't apply optimizations below when no (vector) loop remains, as they all // require one at the moment. VPBasicBlock *HeaderVPBB = vputils::getFirstLoopHeader(*State.Plan, State.VPDT); if (!HeaderVPBB) return; BasicBlock *HeaderBB = State.CFG.VPBB2IRBB[HeaderVPBB]; // Remove redundant induction instructions. legacyCSE(HeaderBB); } 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_range(Uniforms[VF]); 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 getelementptr // instruction contained in the loop. auto IsLoopVaryingGEP = [&](Value *V) { return 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 (!IsLoopVaryingGEP(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) && all_of(I->users(), IsaPred)) 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_range(Uniforms[VF]); // (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 unless the operation is a gather or scatter. // 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 (!IsLoopVaryingGEP(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. bool ScalarInd = 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; // If the induction variable update is a fixed-order recurrence, neither the // induction variable or its update should be marked scalar after // vectorization. auto *IndUpdatePhi = dyn_cast(IndUpdate); if (IndUpdatePhi && Legal->isFixedOrderRecurrence(IndUpdatePhi)) continue; // Determine if all users of the induction variable update instruction are // scalar after vectorization. bool ScalarIndUpdate = 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_range(Worklist); } bool LoopVectorizationCostModel::isScalarWithPredication(Instruction *I, ElementCount VF) { 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; auto *CI = cast(I); // A vector intrinsic or library variant lowering avoids scalarization. return !getVectorIntrinsicIDForCall(CI, TLI) && !hasVectorLibraryVariantFor(*CI, VF, isMaskRequired(CI), TLI); } case Instruction::Load: case Instruction::Store: { bool IsConsecutive = Legal->isConsecutivePtr(getLoadStoreType(I), getLoadStorePointerOperand(I)); return !(IsConsecutive && Config.isLegalMaskedLoadOrStore(I, VF)) && !Config.isLegalGatherOrScatter(I, VF); } case Instruction::UDiv: case Instruction::SDiv: case Instruction::SRem: case Instruction::URem: { // We have the option to use the llvm.masked.udiv intrinsics to avoid // predication. The cost based decision here will always select the masked // intrinsics for scalable vectors as scalarization isn't legal. const auto [ScalarCost, MaskedCost] = getDivRemSpeculationCost(I, VF); return isDivRemScalarWithPredication(ScalarCost, MaskedCost); } } } bool LoopVectorizationCostModel::isMaskRequired(Instruction *I) const { return Legal->isMaskRequired(I, foldTailByMasking()); } // TODO: Fold into LoopVectorizationLegality::isMaskRequired. bool LoopVectorizationCostModel::isPredicatedInst(Instruction *I) const { // TODO: We can use the loop-preheader as context point here and get // context sensitive reasoning for isSafeToSpeculativelyExecute. if (isSafeToSpeculativelyExecute(I) || (isa(I) && !isMaskRequired(I)) || isa(I)) return false; // If the instruction was executed conditionally in the original scalar loop, // predication is needed with a mask whose lanes are all possibly inactive. if (Legal->blockNeedsPredication(I->getParent())) return true; // If we're not folding the tail by masking and not vectorizing a loop with // uncountable exits and side effects, predication is unnecessary. if (!foldTailByMasking() && !Legal->hasUncountableExitWithSideEffects()) return false; // All that remain are instructions with side-effects originally executed in // the loop unconditionally, but now execute under a tail-fold mask (only) // having at least one active lane (the first). If the side-effects of the // instruction are invariant, executing it w/o (the tail-folding) mask is safe // - it will cause the same side-effects as when masked. switch(I->getOpcode()) { default: llvm_unreachable( "instruction should have been considered by earlier checks"); case Instruction::Call: // Side-effects of a Call are assumed to be non-invariant, needing a // (fold-tail) mask. assert(isMaskRequired(I) && "should have returned earlier for calls not needing a mask"); return true; case Instruction::Load: // If the address is loop invariant no predication is needed. return !Legal->isInvariant(getLoadStorePointerOperand(I)); case Instruction::Store: { // 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. return !(Legal->isInvariant(getLoadStorePointerOperand(I)) && TheLoop->isLoopInvariant(cast(I)->getValueOperand())); } case Instruction::UDiv: case Instruction::URem: // If the divisor is loop-invariant no predication is needed. return !Legal->isInvariant(I->getOperand(1)); case Instruction::SDiv: case Instruction::SRem: // Conservative for now, since masked-off lanes may be poison and could // trigger signed overflow. return true; } } uint64_t LoopVectorizationCostModel::getPredBlockCostDivisor( TargetTransformInfo::TargetCostKind CostKind, const BasicBlock *BB) { if (CostKind == TTI::TCK_CodeSize) return 1; // If the block wasn't originally predicated then return early to avoid // computing BlockFrequencyInfo unnecessarily. if (!Legal->blockNeedsPredication(BB)) return 1; uint64_t HeaderFreq = getBFI().getBlockFreq(TheLoop->getHeader()).getFrequency(); uint64_t BBFreq = getBFI().getBlockFreq(BB).getFrequency(); assert(HeaderFreq >= BBFreq && "Header has smaller block freq than dominated BB?"); return std::round((double)HeaderFreq / BBFreq); } static Intrinsic::ID getMaskedDivRemIntrinsic(unsigned Opcode) { switch (Opcode) { case Instruction::UDiv: return Intrinsic::masked_udiv; case Instruction::SDiv: return Intrinsic::masked_sdiv; case Instruction::URem: return Intrinsic::masked_urem; case Instruction::SRem: return Intrinsic::masked_srem; default: llvm_unreachable("Unexpected opcode"); } } std::pair LoopVectorizationCostModel::getDivRemSpeculationCost(Instruction *I, ElementCount VF) { assert(I->getOpcode() == Instruction::UDiv || I->getOpcode() == Instruction::SDiv || I->getOpcode() == Instruction::SRem || I->getOpcode() == Instruction::URem); assert(!isSafeToSpeculativelyExecute(I)); // 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.getFixedValue() * TTI.getCFInstrCost(Instruction::PHI, Config.CostKind); // The cost of the non-predicated instruction. ScalarizationCost += VF.getFixedValue() * TTI.getArithmeticInstrCost( I->getOpcode(), I->getType(), Config.CostKind); // The cost of insertelement and extractelement instructions needed for // scalarization. ScalarizationCost += getScalarizationOverhead(I, VF); // 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 / getPredBlockCostDivisor(Config.CostKind, I->getParent()); } auto *VecTy = toVectorTy(I->getType(), VF); auto *MaskTy = toVectorTy(Type::getInt1Ty(I->getContext()), VF); IntrinsicCostAttributes ICA(getMaskedDivRemIntrinsic(I->getOpcode()), VecTy, {VecTy, VecTy, MaskTy}); InstructionCost MaskedCost = TTI.getIntrinsicInstrCost(ICA, Config.CostKind); return {ScalarizationCost, MaskedCost}; } bool LoopVectorizationCostModel::interleavedAccessCanBeWidened( Instruction *I, ElementCount VF) const { 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."); unsigned InterleaveFactor = Group->getFactor(); // If the instruction's allocated size doesn't equal its type size, it // requires padding and will be scalarized. auto &DL = I->getDataLayout(); auto *ScalarTy = getLoadStoreType(I); if (hasIrregularType(ScalarTy, DL)) return false; // For scalable vectors, the interleave factors must be <= 8 since we require // the (de)interleaveN intrinsics instead of shufflevectors. if (VF.isScalable() && InterleaveFactor > 8) return false; // If the group involves a non-integral pointer, we may not be able to // losslessly cast all values to a common type. bool ScalarNI = DL.isNonIntegralPointerType(ScalarTy); for (Instruction *Member : Group->members()) { 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; 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()) && isMaskRequired(I); bool LoadAccessWithGapsRequiresEpilogMasking = isa(I) && Group->requiresScalarEpilogue() && !isEpilogueAllowed(); bool StoreAccessWithGapsRequiresMasking = isa(I) && !Group->isFull(); 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; // TODO: Support interleaved access that requires a gap mask for scalable VFs. bool NeedsMaskForGaps = LoadAccessWithGapsRequiresEpilogMasking || StoreAccessWithGapsRequiresMasking; if (VF.isScalable() && NeedsMaskForGaps) return false; return Config.isLegalMaskedLoadOrStore(I, VF); } std::optional 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. int Stride = Legal->isConsecutivePtr(ScalarTy, Ptr); if (!Stride) return std::nullopt; // If the instruction is a store located in a predicated block, it will be // scalarized. if (isScalarWithPredication(I, VF)) return std::nullopt; // If the instruction's allocated size doesn't equal it's type size, it // requires padding and will be scalarized. auto &DL = I->getDataLayout(); if (hasIrregularType(ScalarTy, DL)) return std::nullopt; return Stride == 1 ? CM_Widen : CM_Widen_Reverse; } 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 find no uniform value, we won't // analyze again. Uniforms.count(VF) will return 1. Uniforms[VF].clear(); // Now we 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; // Add uniform instructions demanding lane 0 to the worklist. Instructions // that require 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. auto AddToWorklistIfAllowed = [&](Instruction *I) -> void { if (IsOutOfScope(I)) { LLVM_DEBUG(dbgs() << "LV: Found not uniform due to scope: " << *I << "\n"); return; } if (isPredicatedInst(I)) { LLVM_DEBUG( dbgs() << "LV: Found not uniform due to requiring predication: " << *I << "\n"); return; } LLVM_DEBUG(dbgs() << "LV: Found uniform instruction: " << *I << "\n"); Worklist.insert(I); }; // Start with the conditional branches exiting the loop. If the branch // condition is an instruction contained in the loop that is only used by the // branch, it is uniform. Note conditions from uncountable early exits are not // uniform. SmallVector Exiting; TheLoop->getExitingBlocks(Exiting); for (BasicBlock *E : Exiting) { if (Legal->hasUncountableEarlyExit() && TheLoop->getLoopLatch() != E) continue; auto *Cmp = dyn_cast(E->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 choose 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 (!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; } } if (auto *EVI = dyn_cast(&I)) { if (IsOutOfScope(EVI->getAggregateOperand())) { AddToWorklistIfAllowed(EVI); continue; } // Only ExtractValue instructions where the aggregate value comes from a // call are allowed to be non-uniform. assert(isa(EVI->getAggregateOperand()) && "Expected aggregate value to be call return value"); } // If there's no pointer operand, there's nothing to do. auto *Ptr = getLoadStorePointerOperand(&I); if (!Ptr) continue; // If the pointer can be proven to be uniform, always add it to the // worklist. if (isa(Ptr) && isUniform(Ptr, VF)) AddToWorklistIfAllowed(cast(Ptr)); 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); bool UsersAreMemAccesses = all_of(I->users(), [&](User *U) -> bool { auto *UI = cast(U); return TheLoop->contains(UI) && IsVectorizedMemAccessUse(UI, 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. BasicBlock *Latch = TheLoop->getLoopLatch(); 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. bool UniformInd = 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. bool UniformIndUpdate = all_of(IndUpdate->users(), [&](User *U) -> bool { auto *I = cast(U); return I == Ind || 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_range(Worklist); } FixedScalableVFPair LoopVectorizationCostModel::computeMaxVF(ElementCount UserVF, unsigned UserIC) { // Make sure once we return PartialAliasMaskingStatus is not "NotDecided". scope_exit EnsureAliasMaskingStatusIsDecidedOnReturn([this] { if (PartialAliasMaskingStatus == AliasMaskingStatus::NotDecided) PartialAliasMaskingStatus = AliasMaskingStatus::Disabled; }); // For outer loops, use simple type-based heuristic VF. No cost model or // memory dependence analysis is available. if (!TheLoop->isInnermost()) { return Config.computeVPlanOuterloopVF(UserVF); } if (Legal->getRuntimePointerChecking()->Need && TTI.hasBranchDivergence()) { // TODO: It may be 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(); } ScalarEvolution *SE = PSE.getSE(); ElementCount TC = getSmallConstantTripCount(SE, TheLoop); unsigned MaxTC = PSE.getSmallConstantMaxTripCount(); if (!MaxTC && EpilogueLoweringStatus == CM_EpilogueAllowed) MaxTC = getMaxTCFromNonZeroRange(PSE, TheLoop); LLVM_DEBUG(dbgs() << "LV: Found trip count: " << TC << '\n'); if (TC != ElementCount::getFixed(MaxTC)) LLVM_DEBUG(dbgs() << "LV: Found maximum trip count: " << MaxTC << '\n'); if (TC.isScalar()) { reportVectorizationFailure( "Single iteration (non) loop", "loop trip count is one, irrelevant for vectorization", "SingleIterationLoop", ORE, TheLoop); return FixedScalableVFPair::getNone(); } // If BTC matches the widest induction type and is -1 then the trip count // computation will wrap to 0 and the vector trip count will be 0. Do not try // to vectorize. const SCEV *BTC = SE->getBackedgeTakenCount(TheLoop); if (!isa(BTC) && BTC->getType()->getScalarSizeInBits() >= Legal->getWidestInductionType()->getScalarSizeInBits() && SE->isKnownPredicate(CmpInst::ICMP_EQ, BTC, SE->getMinusOne(BTC->getType()))) { reportVectorizationFailure( "Trip count computation wrapped", "backedge-taken count is -1, loop trip count wrapped to 0", "TripCountWrapped", ORE, TheLoop); return FixedScalableVFPair::getNone(); } assert(WideningDecisions.empty() && Uniforms.empty() && Scalars.empty() && "No cost-modeling decisions should have been taken at this point"); switch (EpilogueLoweringStatus) { case CM_EpilogueAllowed: return Config.computeFeasibleMaxVF(MaxTC, UserVF, UserIC, false, requiresScalarEpilogue(true)); case CM_EpilogueNotAllowedFoldTail: [[fallthrough]]; case CM_EpilogueNotNeededFoldTail: LLVM_DEBUG(dbgs() << "LV: tail-folding hint/switch found.\n" << "LV: Not allowing epilogue, creating tail-folded " << "vector loop.\n"); break; case CM_EpilogueNotAllowedLowTripLoop: // fallthrough as a special case of OptForSize case CM_EpilogueNotAllowedOptSize: if (EpilogueLoweringStatus == CM_EpilogueNotAllowedOptSize) LLVM_DEBUG(dbgs() << "LV: Not allowing epilogue due to -Os/-Oz.\n"); else LLVM_DEBUG(dbgs() << "LV: Not allowing epilogue due to low trip " << "count.\n"); // Bail if runtime checks are required, which are not good when optimising // for size. if (Config.runtimeChecksRequired()) return FixedScalableVFPair::getNone(); break; } // Now try the tail folding // Invalidate interleave groups that require an epilogue if we can't mask // the interleave-group. if (!useMaskedInterleavedAccesses(TTI)) { // Note: There is no need to invalidate any cost modeling decisions here, as // none were taken so far (see assertion above). InterleaveInfo.invalidateGroupsRequiringScalarEpilogue(); } FixedScalableVFPair MaxFactors = Config.computeFeasibleMaxVF( MaxTC, UserVF, UserIC, true, requiresScalarEpilogue(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) { MaxPowerOf2RuntimeVF = std::max( *MaxPowerOf2RuntimeVF, *MaxVScale * MaxFactors.ScalableVF.getKnownMinValue()); } else MaxPowerOf2RuntimeVF = std::nullopt; // Stick with tail-folding for now. } auto NoScalarEpilogueNeeded = [this, &UserIC](unsigned MaxVF) { // Return false if the loop is neither a single-latch-exit loop nor an // early-exit loop as tail-folding is not supported in that case. if (TheLoop->getExitingBlock() != TheLoop->getLoopLatch() && !Legal->hasUncountableEarlyExit()) return false; unsigned MaxVFtimesIC = UserIC ? MaxVF * UserIC : MaxVF; ScalarEvolution *SE = PSE.getSE(); // Calling getSymbolicMaxBackedgeTakenCount enables support for loops // with uncountable exits. For countable loops, the symbolic maximum must // remain identical to the known back-edge taken count. const SCEV *BackedgeTakenCount = PSE.getSymbolicMaxBackedgeTakenCount(); assert((Legal->hasUncountableEarlyExit() || BackedgeTakenCount == PSE.getBackedgeTakenCount()) && "Invalid loop count"); const SCEV *ExitCount = SE->getAddExpr( BackedgeTakenCount, SE->getOne(BackedgeTakenCount->getType())); const SCEV *Rem = SE->getURemExpr( SE->applyLoopGuards(ExitCount, TheLoop), SE->getConstant(BackedgeTakenCount->getType(), MaxVFtimesIC)); return Rem->isZero(); }; if (MaxPowerOf2RuntimeVF > 0u) { assert((UserVF.isNonZero() || isPowerOf2_32(*MaxPowerOf2RuntimeVF)) && "MaxFixedVF must be a power of 2"); if (NoScalarEpilogueNeeded(*MaxPowerOf2RuntimeVF)) { // Accept MaxFixedVF if we do not have a tail. LLVM_DEBUG(dbgs() << "LV: No tail will remain for any chosen VF.\n"); return MaxFactors; } } auto ExpectedTC = getSmallBestKnownTC(PSE, TheLoop); if (ExpectedTC && ExpectedTC->isFixed() && ExpectedTC->getFixedValue() <= TTI.getMinTripCountTailFoldingThreshold()) { if (MaxPowerOf2RuntimeVF > 0u) { // If we have a low-trip-count, and the fixed-width VF is known to divide // the trip count but the scalable factor does not, use the fixed-width // factor in preference to allow the generation of a non-predicated loop. if (EpilogueLoweringStatus == CM_EpilogueNotAllowedLowTripLoop && NoScalarEpilogueNeeded(MaxFactors.FixedVF.getFixedValue())) { LLVM_DEBUG(dbgs() << "LV: Picking a fixed-width so that no tail will " "remain for any chosen VF.\n"); MaxFactors.ScalableVF = ElementCount::getScalable(0); return MaxFactors; } } reportVectorizationFailure( "The trip count is below the minial threshold value.", "loop trip count is too low, avoiding vectorization", "LowTripCount", ORE, TheLoop); return FixedScalableVFPair::getNone(); } // 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. bool ContainsScalableVF = MaxFactors.ScalableVF.isNonZero(); setTailFoldingStyle(ContainsScalableVF, UserIC); if (foldTailByMasking()) { if (foldTailWithEVL()) { LLVM_DEBUG( dbgs() << "LV: tail is folded with EVL, forcing unroll factor to be 1. Will " "try to generate VP Intrinsics with scalable vector " "factors only.\n"); // Tail folded loop using VP intrinsics restricts the VF to be scalable // for now. // TODO: extend it for fixed vectors, if required. assert(ContainsScalableVF && "Expected scalable vector factor."); MaxFactors.FixedVF = ElementCount::getFixed(1); } else { tryToEnablePartialAliasMasking(); } return MaxFactors; } // If there was a tail-folding hint/switch, but we can't fold the tail by // masking, fallback to a vectorization with an epilogue. if (EpilogueLoweringStatus == CM_EpilogueNotNeededFoldTail) { LLVM_DEBUG(dbgs() << "LV: Cannot fold tail by masking: vectorize with an " "epilogue instead.\n"); EpilogueLoweringStatus = CM_EpilogueAllowed; return MaxFactors; } if (EpilogueLoweringStatus == CM_EpilogueNotAllowedFoldTail) { LLVM_DEBUG(dbgs() << "LV: Can't fold tail by masking: don't vectorize\n"); return FixedScalableVFPair::getNone(); } if (TC.isZero()) { reportVectorizationFailure( "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(); } void LoopVectorizationPlanner::emitInvalidCostRemarks( OptimizationRemarkEmitter *ORE) { using RecipeVFPair = std::pair; SmallVector InvalidCosts; for (const auto &Plan : VPlans) { for (ElementCount VF : Plan->vectorFactors()) { // The VPlan-based cost model is designed for computing vector cost. // Querying VPlan-based cost model with a scarlar VF will cause some // errors because we expect the VF is vector for most of the widen // recipes. if (VF.isScalar()) continue; VPCostContext CostCtx(CM.TTI, *CM.TLI, *Plan, CM, Config.CostKind, CM.PSE, OrigLoop); precomputeCosts(*Plan, VF, CostCtx); auto Iter = vp_depth_first_deep(Plan->getVectorLoopRegion()->getEntry()); for (VPBasicBlock *VPBB : VPBlockUtils::blocksOnly(Iter)) { for (auto &R : *VPBB) { if (!R.cost(VF, CostCtx).isValid()) InvalidCosts.emplace_back(&R, VF); } } } } if (InvalidCosts.empty()) return; // Emit a report of VFs with invalid costs in the loop. // Group the remarks per recipe, keeping the recipe order from InvalidCosts. DenseMap Numbering; unsigned I = 0; for (auto &Pair : InvalidCosts) if (Numbering.try_emplace(Pair.first, I).second) ++I; // Sort the list, first on recipe(number) then on VF. sort(InvalidCosts, [&Numbering](RecipeVFPair &A, RecipeVFPair &B) { unsigned NA = Numbering[A.first]; unsigned NB = Numbering[B.first]; if (NA != NB) return NA < NB; return ElementCount::isKnownLT(A.second, B.second); }); // For a list of ordered recipe-VF pairs: // [(load, VF1), (load, VF2), (store, VF1)] // group the recipes 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); VPRecipeBase *R = Subset.front().first; unsigned Opcode = TypeSwitch(R) .Case([](const VPHeaderPHIRecipe *R) { return Instruction::PHI; }) .Case( [](const VPWidenStoreRecipe *R) { return Instruction::Store; }) .Case([](const VPWidenLoadRecipe *R) { return Instruction::Load; }) .Case( [](const auto *R) { return Instruction::Call; }) .Case( [](const auto *R) { return R->getOpcode(); }) .Case([](const VPInterleaveRecipe *R) { return R->getStoredValues().empty() ? Instruction::Load : Instruction::Store; }) .Case([](const VPReductionRecipe *R) { return RecurrenceDescriptor::getOpcode(R->getRecurrenceKind()); }); // If the next recipe 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=(VF1, VF2) if (Subset == Tail || Tail[Subset.size()].first != R) { std::string OutString; raw_string_ostream OS(OutString); assert(!Subset.empty() && "Unexpected empty range"); OS << "Recipe with invalid costs prevented vectorization at VF=("; for (const auto &Pair : Subset) OS << (Pair.second == Subset.front().second ? "" : ", ") << Pair.second; OS << "):"; if (Opcode == Instruction::Call) { StringRef Name = ""; if (auto *Int = dyn_cast(R)) { Name = Int->getIntrinsicName(); } else { auto *WidenCall = dyn_cast(R); Function *CalledFn = WidenCall ? WidenCall->getCalledScalarFunction() : cast(R->getOperand(R->getNumOperands() - 1) ->getLiveInIRValue()); Name = CalledFn->getName(); } OS << " call to " << Name; } else OS << " " << Instruction::getOpcodeName(Opcode); reportVectorizationInfo(OutString, "InvalidCost", ORE, OrigLoop, nullptr, R->getDebugLoc()); Tail = Tail.drop_front(Subset.size()); Subset = {}; } else // Grow the subset by one element Subset = Tail.take_front(Subset.size() + 1); } while (!Tail.empty()); } /// Check if any recipe of \p Plan will generate a vector value, which will be /// assigned a vector register. static bool willGenerateVectors(VPlan &Plan, ElementCount VF, const TargetTransformInfo &TTI) { assert(VF.isVector() && "Checking a scalar VF?"); DenseSet EphemeralRecipes; collectEphemeralRecipesForVPlan(Plan, EphemeralRecipes); // Set of already visited types. DenseSet Visited; for (VPBasicBlock *VPBB : VPBlockUtils::blocksOnly( vp_depth_first_shallow(Plan.getVectorLoopRegion()->getEntry()))) { for (VPRecipeBase &R : *VPBB) { if (EphemeralRecipes.contains(&R)) continue; // Continue early if the recipe is considered to not produce a vector // result. Note that this includes VPInstruction where some opcodes may // produce a vector, to preserve existing behavior as VPInstructions model // aspects not directly mapped to existing IR instructions. switch (R.getVPRecipeID()) { case VPRecipeBase::VPDerivedIVSC: case VPRecipeBase::VPScalarIVStepsSC: case VPRecipeBase::VPReplicateSC: case VPRecipeBase::VPInstructionSC: case VPRecipeBase::VPCurrentIterationPHISC: case VPRecipeBase::VPVectorPointerSC: case VPRecipeBase::VPVectorEndPointerSC: case VPRecipeBase::VPExpandSCEVSC: case VPRecipeBase::VPPredInstPHISC: case VPRecipeBase::VPBranchOnMaskSC: continue; case VPRecipeBase::VPReductionSC: case VPRecipeBase::VPActiveLaneMaskPHISC: case VPRecipeBase::VPWidenCallSC: case VPRecipeBase::VPWidenCanonicalIVSC: case VPRecipeBase::VPWidenCastSC: case VPRecipeBase::VPWidenGEPSC: case VPRecipeBase::VPWidenIntrinsicSC: case VPRecipeBase::VPWidenMemIntrinsicSC: case VPRecipeBase::VPWidenSC: case VPRecipeBase::VPBlendSC: case VPRecipeBase::VPFirstOrderRecurrencePHISC: case VPRecipeBase::VPHistogramSC: case VPRecipeBase::VPWidenPHISC: case VPRecipeBase::VPWidenIntOrFpInductionSC: case VPRecipeBase::VPWidenPointerInductionSC: case VPRecipeBase::VPReductionPHISC: case VPRecipeBase::VPInterleaveEVLSC: case VPRecipeBase::VPInterleaveSC: case VPRecipeBase::VPWidenLoadEVLSC: case VPRecipeBase::VPWidenLoadSC: case VPRecipeBase::VPWidenStoreEVLSC: case VPRecipeBase::VPWidenStoreSC: break; default: llvm_unreachable("unhandled recipe"); } auto WillGenerateTargetVectors = [&TTI, VF](Type *VectorTy) { unsigned NumLegalParts = TTI.getNumberOfParts(VectorTy); if (!NumLegalParts) return false; 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. return NumLegalParts <= VF.getKnownMinValue(); } // Two or more elements that share a register - are vectorized. return NumLegalParts < VF.getFixedValue(); }; // If no def nor is a store, e.g., branches, continue - no value to check. if (R.getNumDefinedValues() == 0 && !isa(&R)) continue; // For multi-def recipes, currently only interleaved loads, suffice to // check first def only. // For stores check their stored value; for interleaved stores suffice // the check first stored value only. In all cases this is the second // operand. VPValue *ToCheck = R.getNumDefinedValues() >= 1 ? R.getVPValue(0) : R.getOperand(1); Type *ScalarTy = ToCheck->getScalarType(); if (!Visited.insert({ScalarTy}).second) continue; Type *WideTy = toVectorizedTy(ScalarTy, VF); if (any_of(getContainedTypes(WideTy), WillGenerateTargetVectors)) return true; } } return false; } static bool hasReplicatorRegion(VPlan &Plan) { return any_of(VPBlockUtils::blocksOnly(vp_depth_first_shallow( Plan.getVectorLoopRegion()->getEntry())), [](auto *VPRB) { return VPRB->isReplicator(); }); } /// Returns true if the VPlan contains a VPReductionPHIRecipe with /// FindLast recurrence kind. static bool hasFindLastReductionPhi(VPlan &Plan) { return any_of(Plan.getVectorLoopRegion()->getEntryBasicBlock()->phis(), [](VPRecipeBase &R) { auto *RedPhi = dyn_cast(&R); return RedPhi && RecurrenceDescriptor::isFindLastRecurrenceKind( RedPhi->getRecurrenceKind()); }); } /// Returns true if the VPlan contains header phi recipes that are not currently /// supported for epilogue vectorization. static bool hasUnsupportedHeaderPhiRecipe(VPlan &Plan) { return any_of( Plan.getVectorLoopRegion()->getEntryBasicBlock()->phis(), [](VPRecipeBase &R) { switch (R.getVPRecipeID()) { case VPRecipeBase::VPFirstOrderRecurrencePHISC: // TODO: Add support for fixed-order recurrences. return true; case VPRecipeBase::VPWidenIntOrFpInductionSC: return !cast(&R)->getPHINode(); case VPRecipeBase::VPReductionPHISC: { auto *RedPhi = cast(&R); // TODO: Support FMinNum/FMaxNum, FindLast reductions, and reductions // without underlying values. RecurKind Kind = RedPhi->getRecurrenceKind(); if (RecurrenceDescriptor::isFPMinMaxNumRecurrenceKind(Kind) || RecurrenceDescriptor::isFindLastRecurrenceKind(Kind) || !RedPhi->getUnderlyingValue()) return true; // TODO: Add support for FindIV reductions with sunk expressions: the // resume value from the main loop is in expression domain (e.g., // mul(ReducedIV, 3)), but the epilogue tracks raw IV values. A sunk // expression is identified by a non-VPInstruction user of // ComputeReductionResult. if (RecurrenceDescriptor::isFindIVRecurrenceKind(Kind)) { auto *RdxResult = vputils::findComputeReductionResult(RedPhi); assert(RdxResult && "FindIV reduction must have ComputeReductionResult"); return any_of(RdxResult->users(), std::not_fn(IsaPred)); } return false; } default: return false; }; }); } bool LoopVectorizationPlanner::isCandidateForEpilogueVectorization( VPlan &MainPlan) const { // Bail out if the plan contains header phi recipes not yet supported // for epilogue vectorization. if (hasUnsupportedHeaderPhiRecipe(MainPlan)) 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. // TODO: Add support for loops with an early exit. if (OrigLoop->getExitingBlock() != OrigLoop->getLoopLatch()) return false; return true; } bool LoopVectorizationCostModel::isEpilogueVectorizationProfitable( const ElementCount VF, const unsigned IC) 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. if (!TTI.preferEpilogueVectorization(VF * IC)) return false; unsigned MinVFThreshold = EpilogueVectorizationMinVF.getNumOccurrences() > 0 ? EpilogueVectorizationMinVF : TTI.getEpilogueVectorizationMinVF(); return estimateElementCount(VF * IC, Config.getVScaleForTuning()) >= MinVFThreshold; } std::unique_ptr LoopVectorizationPlanner::selectBestEpiloguePlan( VPlan &MainPlan, ElementCount MainLoopVF, unsigned IC) { if (!EnableEpilogueVectorization) { LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization is disabled.\n"); return nullptr; } if (!CM.isEpilogueAllowed()) { LLVM_DEBUG(dbgs() << "LEV: Unable to vectorize epilogue because no " "epilogue is allowed.\n"); return nullptr; } if (CM.maskPartialAliasing()) { LLVM_DEBUG( dbgs() << "LEV: Epilogue vectorization not supported with alias masking.\n"); return nullptr; } // Not really a cost consideration, but check for unsupported cases here to // simplify the logic. if (!isCandidateForEpilogueVectorization(MainPlan)) { LLVM_DEBUG(dbgs() << "LEV: Unable to vectorize epilogue because the loop " "is not a supported candidate.\n"); return nullptr; } if (hasForcedEpilogueVF()) { if (estimateElementCount(EpilogueVectorizationForceVF, Config.getVScaleForTuning()) >= IC * estimateElementCount(MainLoopVF, Config.getVScaleForTuning())) { // Note that the main loop leaves IC * MainLoopVF iterations iff a scalar // epilogue is required, but then the epilogue loop also requires a scalar // epilogue. LLVM_DEBUG(dbgs() << "LEV: Forced epilogue VF results in dead epilogue " "vector loop, skipping vectorizing epilogue.\n"); return nullptr; } LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization factor is forced.\n"); if (hasPlanWithVF(EpilogueVectorizationForceVF)) { std::unique_ptr Clone( getPlanFor(EpilogueVectorizationForceVF).duplicate()); Clone->setVF(EpilogueVectorizationForceVF); return Clone; } LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization forced factor is not " "viable.\n"); return nullptr; } if (OrigLoop->getHeader()->getParent()->hasOptSize()) { LLVM_DEBUG( dbgs() << "LEV: Epilogue vectorization skipped due to opt for size.\n"); return nullptr; } if (!CM.isEpilogueVectorizationProfitable(MainLoopVF, IC)) { LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization is not profitable for " "this loop\n"); return nullptr; } // Check if a plan's vector loop processes fewer iterations than VF (e.g. when // interleave groups have been narrowed) narrowInterleaveGroups) and return // the adjusted, effective VF. using namespace VPlanPatternMatch; auto GetEffectiveVF = [](VPlan &Plan, ElementCount VF) -> ElementCount { auto *Exiting = Plan.getVectorLoopRegion()->getExitingBasicBlock(); if (match(&Exiting->back(), m_BranchOnCount(m_Add(m_CanonicalIV(), m_Specific(&Plan.getUF())), m_VPValue()))) return ElementCount::get(1, VF.isScalable()); return VF; }; // Check if the main loop processes fewer than MainLoopVF elements per // iteration (e.g. due to narrowing interleave groups). Adjust MainLoopVF // as needed. MainLoopVF = GetEffectiveVF(MainPlan, MainLoopVF); // 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 = ElementCount::getFixed( estimateElementCount(MainLoopVF, Config.getVScaleForTuning())); Type *TCType = Legal->getWidestInductionType(); const SCEV *RemainingIterations = nullptr; unsigned MaxTripCount = 0; const SCEV *TC = vputils::getSCEVExprForVPValue(MainPlan.getTripCount(), PSE); assert(!isa(TC) && "Trip count SCEV must be computable"); const SCEV *KnownMinTC; bool ScalableTC = match(TC, m_scev_c_Mul(m_SCEV(KnownMinTC), m_SCEVVScale())); bool ScalableRemIter = false; ScalarEvolution &SE = *PSE.getSE(); // Use versions of TC and VF in which both are either scalable or fixed. if (ScalableTC == MainLoopVF.isScalable()) { ScalableRemIter = ScalableTC; RemainingIterations = SE.getURemExpr(TC, SE.getElementCount(TCType, MainLoopVF * IC)); } else if (ScalableTC) { const SCEV *EstimatedTC = SE.getMulExpr( KnownMinTC, SE.getConstant(TCType, Config.getVScaleForTuning().value_or(1))); RemainingIterations = SE.getURemExpr( EstimatedTC, SE.getElementCount(TCType, MainLoopVF * IC)); } else RemainingIterations = SE.getURemExpr(TC, SE.getElementCount(TCType, EstimatedRuntimeVF * IC)); // No iterations left to process in the epilogue. if (RemainingIterations->isZero()) return nullptr; if (MainLoopVF.isFixed()) { MaxTripCount = MainLoopVF.getFixedValue() * IC - 1; if (SE.isKnownPredicate(CmpInst::ICMP_ULT, RemainingIterations, SE.getConstant(TCType, MaxTripCount))) { MaxTripCount = SE.getUnsignedRangeMax(RemainingIterations).getZExtValue(); } LLVM_DEBUG(dbgs() << "LEV: Maximum Trip Count for Epilogue: " << MaxTripCount << "\n"); } auto SkipVF = [&](const SCEV *VF, const SCEV *RemIter) -> bool { return SE.isKnownPredicate(CmpInst::ICMP_UGT, VF, RemIter); }; VectorizationFactor Result = VectorizationFactor::Disabled(); VPlan *BestPlan = nullptr; for (auto &NextVF : ProfitableVFs) { // Skip candidate VFs without a corresponding VPlan. if (!hasPlanWithVF(NextVF.Width)) continue; VPlan &CurrentPlan = getPlanFor(NextVF.Width); ElementCount EffectiveVF = GetEffectiveVF(CurrentPlan, NextVF.Width); // Skip candidate VFs with widths >= the (estimated) runtime VF (scalable // vectors) or > the VF of the main loop (fixed vectors). if ((!EffectiveVF.isScalable() && MainLoopVF.isScalable() && ElementCount::isKnownGE(EffectiveVF, EstimatedRuntimeVF)) || (EffectiveVF.isScalable() && ElementCount::isKnownGE(EffectiveVF, MainLoopVF)) || (!EffectiveVF.isScalable() && !MainLoopVF.isScalable() && ElementCount::isKnownGT(EffectiveVF, MainLoopVF))) continue; // If EffectiveVF is greater than the number of remaining iterations, the // epilogue loop would be dead. Skip such factors. If the epilogue plan // also has narrowed interleave groups, use the effective VF since // the epilogue step will be reduced to its IC. // TODO: We should also consider comparing against a scalable // RemainingIterations when SCEV be able to evaluate non-canonical // vscale-based expressions. if (!ScalableRemIter) { // Handle the case where EffectiveVF and RemainingIterations are in // different numerical spaces. if (EffectiveVF.isScalable()) EffectiveVF = ElementCount::getFixed( estimateElementCount(EffectiveVF, Config.getVScaleForTuning())); if (SkipVF(SE.getElementCount(TCType, EffectiveVF), RemainingIterations)) continue; } if (Result.Width.isScalar() || isMoreProfitable(NextVF, Result, MaxTripCount, !hasTailFolded(MainPlan), /*IsEpilogue*/ true)) { Result = NextVF; BestPlan = &CurrentPlan; } } if (!BestPlan) return nullptr; LLVM_DEBUG(dbgs() << "LEV: Vectorizing epilogue loop with VF = " << Result.Width << "\n"); std::unique_ptr Clone(BestPlan->duplicate()); Clone->setVF(Result.Width); return Clone; } unsigned LoopVectorizationPlanner::selectInterleaveCount(VPlan &Plan, 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. // Only interleave tail-folded loops if wide lane masks are requested, as the // overhead of multiple instructions to calculate the predicate is likely // not beneficial. If an epilogue is not allowed for any other reason, // do not interleave. if (!CM.isEpilogueAllowed() && !(CM.preferTailFoldedLoop() && CM.useWideActiveLaneMask())) return 1; if (any_of(Plan.getVectorLoopRegion()->getEntryBasicBlock()->phis(), IsaPred)) { LLVM_DEBUG(dbgs() << "LV: Loop requires variable-length step. " "Unroll factor forced to be 1.\n"); return 1; } // We used the distance for the interleave count. if (!Legal->isSafeForAnyVectorWidth()) return 1; // We don't attempt to perform interleaving for loops with uncountable early // exits because the VPInstruction::AnyOf code cannot currently handle // multiple parts. if (Plan.hasEarlyExit()) return 1; const bool HasReductions = any_of(Plan.getVectorLoopRegion()->getEntryBasicBlock()->phis(), IsaPred); // FIXME: implement interleaving for FindLast transform correctly. if (hasFindLastReductionPhi(Plan)) return 1; VPRegisterUsage R = calculateRegisterUsageForPlan(Plan, {VF}, TTI, CM.ValuesToIgnore)[0]; // 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) { if (VF.isScalar()) LoopCost = CM.expectedCost(VF); else LoopCost = cost(Plan, VF, &R); 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; } // 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 (const 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.contains(Pair.first)) 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. bool HasUnorderedReductions = HasReductions && !any_of(Plan.getVectorLoopRegion()->getEntryBasicBlock()->phis(), [](VPRecipeBase &R) { auto *RedR = dyn_cast(&R); return RedR && RedR->isOrdered(); }); unsigned MaxInterleaveCount = TTI.getMaxInterleaveFactor(VF, HasUnorderedReductions); LLVM_DEBUG(dbgs() << "LV: MaxInterleaveFactor for the target is " << MaxInterleaveCount << "\n"); // Check if the user has overridden the max. if (VF.isScalar()) { if (ForceTargetMaxScalarInterleaveFactor.getNumOccurrences() > 0) MaxInterleaveCount = ForceTargetMaxScalarInterleaveFactor; } else { if (ForceTargetMaxVectorInterleaveFactor.getNumOccurrences() > 0) MaxInterleaveCount = ForceTargetMaxVectorInterleaveFactor; } // Try to get the exact trip count, or an estimate based on profiling data or // ConstantMax from PSE, failing that. auto BestKnownTC = getSmallBestKnownTC(PSE, OrigLoop, /*CanUseConstantMax=*/true, /*CanExcludeZeroTrips=*/CM.isEpilogueAllowed()); // For fixed length VFs treat a scalable trip count as unknown. if (BestKnownTC && (BestKnownTC->isFixed() || VF.isScalable())) { // Re-evaluate trip counts and VFs to be in the same numerical space. unsigned AvailableTC = estimateElementCount(*BestKnownTC, Config.getVScaleForTuning()); unsigned EstimatedVF = estimateElementCount(VF, Config.getVScaleForTuning()); // At least one iteration must be scalar when this constraint holds. So the // maximum available iterations for interleaving is one less. if (requiresScalarEpilogue(Plan, VF)) --AvailableTC; unsigned InterleaveCountLB = bit_floor(std::max( 1u, std::min(AvailableTC / (EstimatedVF * 2), MaxInterleaveCount))); if (getSmallConstantTripCount(PSE.getSE(), OrigLoop).isNonZero()) { // If the best known trip count is exact, 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(AvailableTC / EstimatedVF, MaxInterleaveCount))); MaxInterleaveCount = InterleaveCountLB; if (InterleaveCountUB != InterleaveCountLB) { unsigned TailTripCountUB = (AvailableTC % (EstimatedVF * InterleaveCountUB)); unsigned TailTripCountLB = (AvailableTC % (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 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 = InterleaveCountLB; } } 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 tail-folding 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(OrigLoop->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 AggressivelyInterleave = 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 = 0; unsigned NumLoads = 0; for (VPBasicBlock *VPBB : VPBlockUtils::blocksOnly( vp_depth_first_deep(Plan.getVectorLoopRegion()->getEntry()))) { for (VPRecipeBase &R : *VPBB) { if (isa(&R)) { NumLoads++; continue; } if (isa(&R)) { NumStores++; continue; } if (auto *InterleaveR = dyn_cast(&R)) { if (unsigned StoreOps = InterleaveR->getNumStoreOperands()) NumStores += StoreOps; else NumLoads += InterleaveR->getNumDefinedValues(); continue; } if (auto *RepR = dyn_cast(&R)) { NumLoads += isa(RepR->getUnderlyingInstr()); NumStores += isa(RepR->getUnderlyingInstr()); continue; } if (isa(&R)) { NumLoads++; NumStores++; continue; } } } 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(Plan.getVectorLoopRegion()->getEntryBasicBlock()->phis(), [](VPRecipeBase &R) { auto *RedR = dyn_cast(&R); return RedR && (RecurrenceDescriptor::isAnyOfRecurrenceKind( RedR->getRecurrenceKind()) || RecurrenceDescriptor::isFindIVRecurrenceKind( RedR->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 && OrigLoop->getLoopDepth() > 1) { bool HasOrderedReductions = any_of(Plan.getVectorLoopRegion()->getEntryBasicBlock()->phis(), [](VPRecipeBase &R) { auto *RedR = dyn_cast(&R); return RedR && RedR->isOrdered(); }); if (HasOrderedReductions) { LLVM_DEBUG( dbgs() << "LV: Not interleaving scalar ordered reductions.\n"); return 1; } unsigned F = 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 (VF.isScalar() && AggressivelyInterleave) { 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); } 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 (AggressivelyInterleave) { LLVM_DEBUG(dbgs() << "LV: Interleaving to expose ILP.\n"); return IC; } LLVM_DEBUG(dbgs() << "LV: Not Interleaving.\n"); return 1; } 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) { assert(VF.isVector() && "Expected VF >= 2"); // If we've already collected the instructions to scalarize or the predicated // BBs after vectorization, 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 (InstsToScalarize.contains(VF) || PredicatedBBsAfterVectorization.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]; // 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 logic for: // 1. Scalars after vectorization, as there will only be a single copy // of the instruction. // 2. Scalable VF, as that would lead to invalid scalarization costs. // 3. Emulated masked memrefs, if a hacked cost is needed. if (!isScalarAfterVectorization(&I, VF) && !VF.isScalable() && !useEmulatedMaskMemRefHack(&I, VF) && computePredInstDiscount(&I, ScalarCosts, VF) >= 0) { for (const auto &[I, IC] : ScalarCosts) ScalarCostsVF.insert({I, IC}); } // Remember that BB will remain after vectorization. PredicatedBBsAfterVectorization[VF].insert(BB); for (auto *Pred : predecessors(BB)) { if (Pred->getSingleSuccessor() == BB) PredicatedBBsAfterVectorization[VF].insert(Pred); } } } } 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; // Cannot scalarize fixed-order recurrence phis at the moment. if (isa(I) && Legal->isFixedOrderRecurrence(cast(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); // 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)); // Compute the scalarization overhead of needed insertelement instructions // and phi nodes. if (isScalarWithPredication(I, VF) && !I->getType()->isVoidTy()) { Type *WideTy = toVectorizedTy(I->getType(), VF); for (Type *VectorTy : getContainedTypes(WideTy)) { ScalarCost += TTI.getScalarizationOverhead( cast(VectorTy), APInt::getAllOnes(VF.getFixedValue()), /*Insert=*/true, /*Extract=*/false, Config.CostKind); } ScalarCost += VF.getFixedValue() * TTI.getCFInstrCost(Instruction::PHI, Config.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(canVectorizeTy(J->getType()) && "Instruction has non-scalar type"); if (CanBeScalarized(J)) Worklist.push_back(J); else if (needsExtract(J, VF)) { Type *WideTy = toVectorizedTy(J->getType(), VF); for (Type *VectorTy : getContainedTypes(WideTy)) { ScalarCost += TTI.getScalarizationOverhead( cast(VectorTy), APInt::getAllOnes(VF.getFixedValue()), /*Insert*/ false, /*Extract*/ true, Config.CostKind); } } } // Scale the total scalar cost by block probability. ScalarCost /= getPredBlockCostDivisor(Config.CostKind, I->getParent()); // 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; } InstructionCost LoopVectorizationCostModel::expectedCost(ElementCount VF) { InstructionCost Cost; assert(VF.isScalar() && "must only be called for scalar VFs"); // For each block. for (BasicBlock *BB : TheLoop->blocks()) { InstructionCost BlockCost; // For each instruction in the old loop. for (Instruction &I : *BB) { // Skip ignored values. if (ValuesToIgnore.count(&I) || (VF.isVector() && VecValuesToIgnore.count(&I))) continue; InstructionCost C = getInstructionCost(&I, VF); // Check if we should override the cost. if (C.isValid() && ForceTargetInstructionCost.getNumOccurrences() > 0) C = InstructionCost(ForceTargetInstructionCost); BlockCost += C; LLVM_DEBUG(dbgs() << "LV: Found an estimated cost of " << C << " for VF " << VF << " For instruction: " << I << '\n'); } // In the scalar loop, 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. getPredBlockCostDivisor will return 1 for blocks that are // only predicated by the header mask when folding the tail. Cost += BlockCost / getPredBlockCostDivisor(Config.CostKind, BB); } return Cost; } /// Gets the address access SCEV for Ptr, if it should be used for cost modeling /// according to isAddressSCEVForCost. /// /// This SCEV can be sent to the Target in order to estimate the address /// calculation cost. static const SCEV *getAddressAccessSCEV( Value *Ptr, PredicatedScalarEvolution &PSE, const Loop *TheLoop) { const SCEV *Addr = PSE.getSCEV(Ptr); return vputils::isAddressSCEVForCost(Addr, *PSE.getSE(), TheLoop) ? Addr : nullptr; } 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, PSE, TheLoop); // Get the cost of the scalar memory instruction and address computation. InstructionCost Cost = VF.getFixedValue() * TTI.getAddressComputationCost(PtrTy, SE, PtrSCEV, Config.CostKind); // 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. const Align Alignment = getLoadStoreAlignment(I); TTI::OperandValueInfo OpInfo = TTI::getOperandInfo(I->getOperand(0)); Cost += VF.getFixedValue() * TTI.getMemoryOpCost(I->getOpcode(), ValTy->getScalarType(), Alignment, AS, Config.CostKind, OpInfo); // Get the overhead of the extractelement and insertelement instructions // we might create due to scalarization. Cost += getScalarizationOverhead(I, VF); // 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 /= getPredBlockCostDivisor(Config.CostKind, I->getParent()); // Add the cost of an i1 extract and a branch auto *VecI1Ty = VectorType::get(IntegerType::getInt1Ty(ValTy->getContext()), VF); Cost += TTI.getScalarizationOverhead( VecI1Ty, APInt::getAllOnes(VF.getFixedValue()), /*Insert=*/false, /*Extract=*/true, Config.CostKind); Cost += TTI.getCFInstrCost(Instruction::CondBr, Config.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, InstWidening Kind) { assert((Kind == CM_Widen || Kind == CM_Widen_Reverse) && "Expected a consecutive widening decision"); Type *ValTy = getLoadStoreType(I); auto *VectorTy = cast(toVectorTy(ValTy, VF)); unsigned AS = getLoadStoreAddressSpace(I); const Align Alignment = getLoadStoreAlignment(I); InstructionCost Cost = 0; if (isMaskRequired(I)) { unsigned IID = I->getOpcode() == Instruction::Load ? Intrinsic::masked_load : Intrinsic::masked_store; Cost += TTI.getMemIntrinsicInstrCost( MemIntrinsicCostAttributes(IID, VectorTy, Alignment, AS), Config.CostKind); } else { TTI::OperandValueInfo OpInfo = TTI::getOperandInfo(I->getOperand(0)); Cost += TTI.getMemoryOpCost(I->getOpcode(), VectorTy, Alignment, AS, Config.CostKind, OpInfo, I); } if (Kind == CM_Widen_Reverse) Cost += TTI.getShuffleCost(TargetTransformInfo::SK_Reverse, VectorTy, VectorTy, {}, Config.CostKind, 0); return Cost; } InstructionCost LoopVectorizationCostModel::getUniformMemOpCost(Instruction *I, ElementCount VF) { assert(isUniformMemOp(*I, VF)); Type *ValTy = getLoadStoreType(I); Type *PtrTy = getLoadStorePointerOperand(I)->getType(); auto *VectorTy = cast(toVectorTy(ValTy, VF)); const Align Alignment = getLoadStoreAlignment(I); unsigned AS = getLoadStoreAddressSpace(I); if (isa(I)) { return TTI.getAddressComputationCost(PtrTy, nullptr, nullptr, Config.CostKind) + TTI.getMemoryOpCost(Instruction::Load, ValTy, Alignment, AS, Config.CostKind) + TTI.getShuffleCost(TargetTransformInfo::SK_Broadcast, VectorTy, VectorTy, {}, Config.CostKind); } StoreInst *SI = cast(I); bool IsLoopInvariantStoreValue = Legal->isInvariant(SI->getValueOperand()); // TODO: We have existing tests that request the cost of extracting element // VF.getKnownMinValue() - 1 from a scalable vector. This does not represent // the actual generated code, which involves extracting the last element of // a scalable vector where the lane to extract is unknown at compile time. InstructionCost Cost = TTI.getAddressComputationCost(PtrTy, nullptr, nullptr, Config.CostKind) + TTI.getMemoryOpCost(Instruction::Store, ValTy, Alignment, AS, Config.CostKind); if (!IsLoopInvariantStoreValue) Cost += TTI.getIndexedVectorInstrCostFromEnd(Instruction::ExtractElement, VectorTy, Config.CostKind, 0); return Cost; } InstructionCost LoopVectorizationCostModel::getGatherScatterCost(Instruction *I, ElementCount VF) { Type *ValTy = getLoadStoreType(I); auto *VectorTy = cast(toVectorTy(ValTy, VF)); const Align Alignment = getLoadStoreAlignment(I); Value *Ptr = getLoadStorePointerOperand(I); Type *PtrTy = Ptr->getType(); if (!isUniform(Ptr, VF)) PtrTy = toVectorTy(PtrTy, VF); unsigned IID = I->getOpcode() == Instruction::Load ? Intrinsic::masked_gather : Intrinsic::masked_scatter; return TTI.getAddressComputationCost(PtrTy, nullptr, nullptr, Config.CostKind) + TTI.getMemIntrinsicInstrCost( MemIntrinsicCostAttributes(IID, VectorTy, Ptr, isMaskRequired(I), Alignment, I), Config.CostKind); } InstructionCost LoopVectorizationCostModel::getInterleaveGroupCost(Instruction *I, ElementCount VF) { const auto *Group = getInterleavedAccessGroup(I); assert(Group && "Fail to get an interleaved access group."); Instruction *InsertPos = Group->getInsertPos(); Type *ValTy = getLoadStoreType(InsertPos); auto *VectorTy = cast(toVectorTy(ValTy, VF)); unsigned AS = getLoadStoreAddressSpace(InsertPos); 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() && !isEpilogueAllowed()) || (isa(I) && !Group->isFull()); InstructionCost Cost = TTI.getInterleavedMemoryOpCost( InsertPos->getOpcode(), WideVecTy, Group->getFactor(), Indices, Group->getAlign(), AS, Config.CostKind, isMaskRequired(I), UseMaskForGaps); if (Group->isReverse()) { // TODO: Add support for reversed masked interleaved access. assert(!isMaskRequired(I) && "Reverse masked interleaved access not supported."); Cost += Group->getNumMembers() * TTI.getShuffleCost(TargetTransformInfo::SK_Reverse, VectorTy, VectorTy, {}, Config.CostKind, 0); } return Cost; } std::optional LoopVectorizationCostModel::getReductionPatternCost(Instruction *I, ElementCount VF, Type *Ty) const { using namespace llvm::PatternMatch; // Early exit for no inloop reductions if (Config.getInLoopReductions().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. Instruction *LastChain = Config.getInLoopReductionImmediateChain(RetI); if (!LastChain) return std::nullopt; // Find the reduction this chain is a part of and calculate the basic cost of // the reduction on its own. Instruction *ReductionPhi = LastChain; while (!isa(ReductionPhi)) ReductionPhi = Config.getInLoopReductionImmediateChain(ReductionPhi); const RecurrenceDescriptor &RdxDesc = Legal->getRecurrenceDescriptor(cast(ReductionPhi)); InstructionCost BaseCost; RecurKind RK = RdxDesc.getRecurrenceKind(); if (RecurrenceDescriptor::isMinMaxRecurrenceKind(RK)) { Intrinsic::ID MinMaxID = getMinMaxReductionIntrinsicOp(RK); BaseCost = TTI.getMinMaxReductionCost( MinMaxID, VectorTy, RdxDesc.getFastMathFlags(), Config.CostKind); } else { BaseCost = TTI.getArithmeticReductionCost(RdxDesc.getOpcode(), VectorTy, RdxDesc.getFastMathFlags(), Config.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 (RK == RecurKind::FMulAdd) BaseCost += TTI.getArithmeticInstrCost(Instruction::FMul, VectorTy, Config.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 (Config.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, Config.CostKind, Op0); InstructionCost MulCost = TTI.getArithmeticInstrCost(Instruction::Mul, MulType, Config.CostKind); InstructionCost Ext2Cost = TTI.getCastInstrCost( RedOp->getOpcode(), VectorTy, MulType, TTI::CastContextHint::None, Config.CostKind, RedOp); InstructionCost RedCost = TTI.getMulAccReductionCost( IsUnsigned, RdxDesc.getOpcode(), RdxDesc.getRecurrenceType(), ExtType, Config.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(), Config.CostKind); InstructionCost ExtCost = TTI.getCastInstrCost( RedOp->getOpcode(), VectorTy, ExtType, TTI::CastContextHint::None, Config.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, Config.CostKind, Op0); InstructionCost ExtCost1 = TTI.getCastInstrCost( Op1->getOpcode(), VectorTy, VectorType::get(Op1Ty, VectorTy), TTI::CastContextHint::None, Config.CostKind, Op1); InstructionCost MulCost = TTI.getArithmeticInstrCost( Instruction::Mul, VectorTy, Config.CostKind); InstructionCost RedCost = TTI.getMulAccReductionCost( IsUnsigned, RdxDesc.getOpcode(), RdxDesc.getRecurrenceType(), ExtType, Config.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, Config.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, Config.CostKind); InstructionCost RedCost = TTI.getMulAccReductionCost( true, RdxDesc.getOpcode(), RdxDesc.getRecurrenceType(), VectorTy, Config.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); Type *PtrTy = getLoadStorePointerOperand(I)->getType(); const Align Alignment = getLoadStoreAlignment(I); unsigned AS = getLoadStoreAddressSpace(I); TTI::OperandValueInfo OpInfo = TTI::getOperandInfo(I->getOperand(0)); return TTI.getAddressComputationCost(PtrTy, nullptr, nullptr, Config.CostKind) + TTI.getMemoryOpCost(I->getOpcode(), ValTy, Alignment, AS, Config.CostKind, OpInfo, I); } return getWideningCost(I, VF); } InstructionCost LoopVectorizationCostModel::getScalarizationOverhead(Instruction *I, ElementCount VF) 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 = toVectorizedTy(I->getType(), VF); if (!RetTy->isVoidTy() && (!isa(I) || !TTI.supportsEfficientVectorElementLoadStore())) { TTI::VectorInstrContext VIC = TTI::VectorInstrContext::None; if (isa(I)) VIC = TTI::VectorInstrContext::Load; else if (isa(I)) VIC = TTI::VectorInstrContext::Store; for (Type *VectorTy : getContainedTypes(RetTy)) { Cost += TTI.getScalarizationOverhead( cast(VectorTy), APInt::getAllOnes(VF.getFixedValue()), /*Insert=*/true, /*Extract=*/false, Config.CostKind, /*ForPoisonSrc=*/true, {}, VIC); } } // 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)); TTI::VectorInstrContext OperandVIC = isa(I) ? TTI::VectorInstrContext::Store : TTI::VectorInstrContext::None; return Cost + TTI.getOperandsScalarizationOverhead(Tys, Config.CostKind, OperandVIC); } void LoopVectorizationCostModel::setCostBasedWideningDecision(ElementCount VF) { if (VF.isScalar()) return; // 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). NumPredStores = 0; for (BasicBlock *BB : TheLoop->blocks()) for (Instruction &I : *BB) if (isa(&I) && isScalarWithPredication(&I, VF)) ++NumPredStores; for (BasicBlock *BB : TheLoop->blocks()) { // For each instruction in the old loop. for (Instruction &I : *BB) { Value *Ptr = getLoadStorePointerOperand(&I); if (!Ptr) continue; if (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 = Config.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 (std::optional Decision = memoryInstructionCanBeWidened(&I, VF)) { setWideningDecision(&I, VF, *Decision, getConsecutiveMemOpCost(&I, VF, *Decision)); continue; } // Choose between Interleaving, Gather/Scatter or Scalarization. InstructionCost InterleaveCost = InstructionCost::getInvalid(); unsigned NumAccesses = 1; if (isAccessInterleaved(&I)) { const 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 = Config.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 (const auto *Group = getInterleavedAccessGroup(&I)) { if (Decision == CM_Scalarize) { for (Instruction *I : Group->members()) setWideningDecision(I, VF, Decision, getMemInstScalarizationCost(I, VF)); } else { 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. SmallSetVector 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 (TheLoop->contains(InstOp) && !isa(InstOp) && AddrDefs.insert(InstOp)) Worklist.push_back(InstOp); } auto UpdateMemOpUserCost = [this, VF](LoadInst *LI) { // If there are direct memory op users of the newly scalarized load, // their cost may have changed because there's no scalarization // overhead for the operand. Update it. for (User *U : LI->users()) { if (!isa(U)) continue; if (getWideningDecision(cast(U), VF) != CM_Scalarize) continue; setWideningDecision( cast(U), VF, CM_Scalarize, getMemInstScalarizationCost(cast(U), VF)); } }; 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 (!isPredicatedInst(I) && (Decision == CM_Widen || Decision == CM_Widen_Reverse || (!isUniformMemOp(*I, VF) && Decision == CM_Scalarize))) { // Scalarize a widened load of address or update the cost of a scalar // load of an address. setWideningDecision( I, VF, CM_Scalarize, (VF.getKnownMinValue() * getMemoryInstructionCost(I, ElementCount::getFixed(1)))); UpdateMemOpUserCost(cast(I)); } else if (const auto *Group = getInterleavedAccessGroup(I)) { // Scalarize all members of this interleaved group when any member // is used as an address. The address-used load skips scalarization // overhead, other members include it. for (Instruction *Member : Group->members()) { InstructionCost Cost = AddrDefs.contains(Member) ? (VF.getKnownMinValue() * getMemoryInstructionCost( Member, ElementCount::getFixed(1))) : getMemInstScalarizationCost(Member, VF); setWideningDecision(Member, VF, CM_Scalarize, Cost); UpdateMemOpUserCost(cast(Member)); } } } else { // Cannot scalarize fixed-order recurrence phis at the moment. if (isa(I) && Legal->isFixedOrderRecurrence(cast(I))) continue; // Make sure I gets scalarized and a cost estimate without // scalarization overhead. ForcedScalars[VF].insert(I); } } } bool LoopVectorizationCostModel::shouldConsiderInvariant(Value *Op) { if (!Legal->isInvariant(Op)) return false; // Consider Op invariant, if it or its operands aren't predicated // instruction in the loop. In that case, it is not trivially hoistable. auto *OpI = dyn_cast(Op); return !OpI || !TheLoop->contains(OpI) || (!isPredicatedInst(OpI) && (!isa(OpI) || OpI->getParent() != TheLoop->getHeader()) && all_of(OpI->operands(), [this](Value *Op) { return shouldConsiderInvariant(Op); })); } InstructionCost 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 InstsToScalarize[VF][I]; // 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 getInstructionCost(I, ElementCount::getFixed(1)) * VF.getKnownMinValue(); } const auto &MinBWs = Config.getMinimalBitwidths(); uint64_t InstrMinBWs = MinBWs.lookup(I); Type *RetTy = I->getType(); if (canTruncateToMinimalBitwidth(I, VF)) RetTy = IntegerType::get(RetTy->getContext(), InstrMinBWs); auto *SE = PSE.getSE(); Type *VectorTy; if (isScalarAfterVectorization(I, VF)) { [[maybe_unused]] 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); }); }; // 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 = toVectorizedTy(RetTy, VF); if (VF.isVector() && VectorTy->isVectorTy() && !TTI.getNumberOfParts(VectorTy)) return InstructionCost::getInvalid(); // 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::UncondBr: case Instruction::CondBr: { // 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. // Note that the conditional branch from the loop latch will be replaced by // a single branch controlling the loop, so there is no extra overhead from // scalarization. bool ScalarPredicatedBB = false; CondBrInst *BI = dyn_cast(I); if (VF.isVector() && BI && (PredicatedBBsAfterVectorization[VF].count(BI->getSuccessor(0)) || PredicatedBBsAfterVectorization[VF].count(BI->getSuccessor(1))) && BI->getParent() != TheLoop->getLoopLatch()) 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 *VecI1Ty = VectorType::get(IntegerType::getInt1Ty(RetTy->getContext()), VF); return (TTI.getScalarizationOverhead( VecI1Ty, APInt::getAllOnes(VF.getFixedValue()), /*Insert*/ false, /*Extract*/ true, Config.CostKind) + (TTI.getCFInstrCost(Instruction::CondBr, Config.CostKind) * VF.getFixedValue())); } if (I->getParent() == TheLoop->getLoopLatch() || VF.isScalar()) // The back-edge branch will remain, as will all scalar branches. return TTI.getCFInstrCost(Instruction::UncondBr, Config.CostKind); // 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::Switch: { if (VF.isScalar()) return TTI.getCFInstrCost(Instruction::Switch, Config.CostKind); auto *Switch = cast(I); return Switch->getNumCases() * TTI.getCmpSelInstrCost( Instruction::ICmp, toVectorTy(Switch->getCondition()->getType(), VF), toVectorTy(Type::getInt1Ty(I->getContext()), VF), CmpInst::ICMP_EQ, Config.CostKind); } case Instruction::PHI: { auto *Phi = cast(I); // First-order recurrences are replaced by vector shuffles inside the loop. if (VF.isVector() && Legal->isFixedOrderRecurrence(Phi)) { return TTI.getShuffleCost( TargetTransformInfo::SK_Splice, cast(VectorTy), cast(VectorTy), {}, Config.CostKind, -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()) { Type *ResultTy = Phi->getType(); // All instructions in an Any-of reduction chain are narrowed to bool. // Check if that is the case for this phi node. auto *HeaderUser = cast_if_present( find_singleton(Phi->users(), [this](User *U, bool) -> User * { auto *Phi = dyn_cast(U); if (Phi && Phi->getParent() == TheLoop->getHeader()) return Phi; return nullptr; })); if (HeaderUser) { auto &ReductionVars = Legal->getReductionVars(); auto Iter = ReductionVars.find(HeaderUser); if (Iter != ReductionVars.end() && RecurrenceDescriptor::isAnyOfRecurrenceKind( Iter->second.getRecurrenceKind())) ResultTy = Type::getInt1Ty(Phi->getContext()); } return (Phi->getNumIncomingValues() - 1) * TTI.getCmpSelInstrCost( Instruction::Select, toVectorTy(ResultTy, VF), toVectorTy(Type::getInt1Ty(Phi->getContext()), VF), CmpInst::BAD_ICMP_PREDICATE, Config.CostKind); } // When tail folding with EVL, if the phi is part of an out of loop // reduction then it will be transformed into a wide vp_merge. if (VF.isVector() && foldTailWithEVL() && Legal->getReductionVars().contains(Phi) && !Config.isInLoopReduction(Phi)) { IntrinsicCostAttributes ICA( Intrinsic::vp_merge, toVectorTy(Phi->getType(), VF), {toVectorTy(Type::getInt1Ty(Phi->getContext()), VF)}); return TTI.getIntrinsicInstrCost(ICA, Config.CostKind); } return TTI.getCFInstrCost(Instruction::PHI, Config.CostKind); } case Instruction::UDiv: case Instruction::SDiv: case Instruction::URem: case Instruction::SRem: if (VF.isVector() && isPredicatedInst(I)) { const auto [ScalarCost, MaskedCost] = getDivRemSpeculationCost(I, VF); return isDivRemScalarWithPredication(ScalarCost, MaskedCost) ? ScalarCost : MaskedCost; } // We've proven all lanes safe to speculate, fall through. [[fallthrough]]; case Instruction::Add: case Instruction::Sub: { auto Info = Legal->getHistogramInfo(I); if (Info && VF.isVector()) { const HistogramInfo *HGram = Info.value(); // Assume that a non-constant update value (or a constant != 1) requires // a multiply, and add that into the cost. InstructionCost MulCost = TTI::TCC_Free; ConstantInt *RHS = dyn_cast(I->getOperand(1)); if (!RHS || RHS->getZExtValue() != 1) MulCost = TTI.getArithmeticInstrCost(Instruction::Mul, VectorTy, Config.CostKind); // Find the cost of the histogram operation itself. Type *PtrTy = VectorType::get(HGram->Load->getPointerOperandType(), VF); Type *ScalarTy = I->getType(); Type *MaskTy = VectorType::get(Type::getInt1Ty(I->getContext()), VF); IntrinsicCostAttributes ICA(Intrinsic::experimental_vector_histogram_add, Type::getVoidTy(I->getContext()), {PtrTy, ScalarTy, MaskTy}); // Add the costs together with the add/sub operation. return TTI.getIntrinsicInstrCost(ICA, Config.CostKind) + MulCost + TTI.getArithmeticInstrCost(I->getOpcode(), VectorTy, Config.CostKind); } [[fallthrough]]; } case Instruction::FAdd: 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 && ((TheLoop->isLoopInvariant(I->getOperand(0)) && PSE.getSCEV(I->getOperand(0))->isOne()) || (TheLoop->isLoopInvariant(I->getOperand(1)) && PSE.getSCEV(I->getOperand(1))->isOne()))) return 0; // Detect reduction patterns if (auto RedCost = getReductionPatternCost(I, VF, VectorTy)) 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); if (!isa(Op2) && TheLoop->isLoopInvariant(Op2) && PSE.getSE()->isSCEVable(Op2->getType()) && isa(PSE.getSCEV(Op2))) { Op2 = cast(PSE.getSCEV(Op2))->getValue(); } auto Op2Info = TTI.getOperandInfo(Op2); if (Op2Info.Kind == TargetTransformInfo::OK_AnyValue && shouldConsiderInvariant(Op2)) Op2Info.Kind = TargetTransformInfo::OK_UniformValue; SmallVector Operands(I->operand_values()); return TTI.getArithmeticInstrCost( I->getOpcode(), VectorTy, Config.CostKind, {TargetTransformInfo::OK_AnyValue, TargetTransformInfo::OP_None}, Op2Info, Operands, I, TLI); } case Instruction::FNeg: { return TTI.getArithmeticInstrCost( I->getOpcode(), VectorTy, Config.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); return TTI.getArithmeticInstrCost( match(I, m_LogicalOr()) ? Instruction::Or : Instruction::And, VectorTy, Config.CostKind, {Op1VK, Op1VP}, {Op2VK, Op2VP}, {Op0, Op1}, 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, Config.CostKind, {TTI::OK_AnyValue, TTI::OP_None}, {TTI::OK_AnyValue, TTI::OP_None}, I); } case Instruction::ICmp: case Instruction::FCmp: { Type *ValTy = I->getOperand(0)->getType(); if (canTruncateToMinimalBitwidth(I, VF)) { [[maybe_unused]] Instruction *Op0AsInstruction = dyn_cast(I->getOperand(0)); assert((!canTruncateToMinimalBitwidth(Op0AsInstruction, VF) || InstrMinBWs == MinBWs.lookup(Op0AsInstruction)) && "if both the operand and the compare are marked for " "truncation, they must have the same bitwidth"); ValTy = IntegerType::get(ValTy->getContext(), InstrMinBWs); } VectorTy = toVectorTy(ValTy, VF); return TTI.getCmpSelInstrCost( I->getOpcode(), VectorTy, CmpInst::makeCmpResultType(VectorTy), cast(I)->getPredicate(), Config.CostKind, {TTI::OK_AnyValue, TTI::OP_None}, {TTI::OK_AnyValue, TTI::OP_None}, 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 isPredicatedInst(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_InvalidatedDecision: return TTI::CastContextHint::None; } 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, Config.CostKind, Trunc); } // Detect reduction patterns if (auto RedCost = getReductionPatternCost(I, VF, VectorTy)) return *RedCost; Type *SrcScalarTy = I->getOperand(0)->getType(); Instruction *Op0AsInstruction = dyn_cast(I->getOperand(0)); if (canTruncateToMinimalBitwidth(Op0AsInstruction, VF)) SrcScalarTy = IntegerType::get(SrcScalarTy->getContext(), MinBWs.lookup(Op0AsInstruction)); Type *SrcVecTy = VectorTy->isVectorTy() ? toVectorTy(SrcScalarTy, VF) : SrcScalarTy; if (canTruncateToMinimalBitwidth(I, VF)) { // If the result type is <= the source type, there will be no extend // after truncating the users to the minimal required bitwidth. if (VectorTy->getScalarSizeInBits() <= SrcVecTy->getScalarSizeInBits() && (I->getOpcode() == Instruction::ZExt || I->getOpcode() == Instruction::SExt)) return 0; } return TTI.getCastInstrCost(Opcode, VectorTy, SrcVecTy, CCH, Config.CostKind, I); } case Instruction::Call: return getVectorCallCost(cast(I), VF); case Instruction::ExtractValue: return TTI.getInstructionCost(I, Config.CostKind); 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(); return TTI.getArithmeticInstrCost(Instruction::Mul, RetTy, Config.CostKind); case Instruction::Freeze: return TTI::TCC_Free; default: // This opcode is unknown. Assume that it is the same as 'mul'. return TTI.getArithmeticInstrCost(Instruction::Mul, VectorTy, Config.CostKind); } // end of switch. } void LoopVectorizationCostModel::collectValuesToIgnore() { // Ignore ephemeral values. CodeMetrics::collectEphemeralValues(TheLoop, AC, ValuesToIgnore); SmallVector DeadInterleavePointerOps; SmallVector DeadOps; // If a scalar epilogue is required, users outside the loop won't use // live-outs from the vector loop but from the scalar epilogue. Ignore them if // that is the case. bool RequiresScalarEpilogue = requiresScalarEpilogue(true); auto IsLiveOutDead = [this, RequiresScalarEpilogue](User *U) { return RequiresScalarEpilogue && !TheLoop->contains(cast(U)->getParent()); }; LoopBlocksDFS DFS(TheLoop); DFS.perform(LI); for (BasicBlock *BB : reverse(make_range(DFS.beginRPO(), DFS.endRPO()))) for (Instruction &I : reverse(*BB)) { if (VecValuesToIgnore.contains(&I) || ValuesToIgnore.contains(&I)) continue; // Add instructions that would be trivially dead and are only used by // values already ignored to DeadOps to seed worklist. if (wouldInstructionBeTriviallyDead(&I, TLI) && all_of(I.users(), [this, IsLiveOutDead](User *U) { return VecValuesToIgnore.contains(U) || ValuesToIgnore.contains(U) || IsLiveOutDead(U); })) DeadOps.push_back(&I); // For interleave groups, we only create a pointer for the start of the // interleave group. Queue up addresses of group members except the insert // position for further processing. if (isAccessInterleaved(&I)) { auto *Group = getInterleavedAccessGroup(&I); if (Group->getInsertPos() == &I) continue; Value *PointerOp = getLoadStorePointerOperand(&I); DeadInterleavePointerOps.push_back(PointerOp); } // Queue branches for analysis. They are dead, if their successors only // contain dead instructions. if (isa(&I)) DeadOps.push_back(&I); } // Mark ops feeding interleave group members as free, if they are only used // by other dead computations. for (unsigned I = 0; I != DeadInterleavePointerOps.size(); ++I) { auto *Op = dyn_cast(DeadInterleavePointerOps[I]); if (!Op || !TheLoop->contains(Op) || any_of(Op->users(), [this](User *U) { Instruction *UI = cast(U); return !VecValuesToIgnore.contains(U) && (!isAccessInterleaved(UI) || getInterleavedAccessGroup(UI)->getInsertPos() == UI); })) continue; VecValuesToIgnore.insert(Op); append_range(DeadInterleavePointerOps, Op->operands()); } // Mark ops that would be trivially dead and are only used by ignored // instructions as free. BasicBlock *Header = TheLoop->getHeader(); // Returns true if the block contains only dead instructions. Such blocks will // be removed by VPlan-to-VPlan transforms and won't be considered by the // VPlan-based cost model, so skip them in the legacy cost-model as well. auto IsEmptyBlock = [this](BasicBlock *BB) { return all_of(*BB, [this](Instruction &I) { return ValuesToIgnore.contains(&I) || VecValuesToIgnore.contains(&I) || isa(&I); }); }; for (unsigned I = 0; I != DeadOps.size(); ++I) { auto *Op = dyn_cast(DeadOps[I]); // Check if the branch should be considered dead. if (auto *Br = dyn_cast_or_null(Op)) { BasicBlock *ThenBB = Br->getSuccessor(0); BasicBlock *ElseBB = Br->getSuccessor(1); // Don't considers branches leaving the loop for simplification. if (!TheLoop->contains(ThenBB) || !TheLoop->contains(ElseBB)) continue; bool ThenEmpty = IsEmptyBlock(ThenBB); bool ElseEmpty = IsEmptyBlock(ElseBB); if ((ThenEmpty && ElseEmpty) || (ThenEmpty && ThenBB->getSingleSuccessor() == ElseBB && ElseBB->phis().empty()) || (ElseEmpty && ElseBB->getSingleSuccessor() == ThenBB && ThenBB->phis().empty())) { VecValuesToIgnore.insert(Br); DeadOps.push_back(Br->getCondition()); } continue; } // Skip any op that shouldn't be considered dead. if (!Op || !TheLoop->contains(Op) || (isa(Op) && Op->getParent() == Header) || !wouldInstructionBeTriviallyDead(Op, TLI) || any_of(Op->users(), [this, IsLiveOutDead](User *U) { return !VecValuesToIgnore.contains(U) && !ValuesToIgnore.contains(U) && !IsLiveOutDead(U); })) continue; // If all of Op's users are in ValuesToIgnore, add it to ValuesToIgnore // which applies for both scalar and vector versions. Otherwise it is only // dead in vector versions, so only add it to VecValuesToIgnore. if (all_of(Op->users(), [this](User *U) { return ValuesToIgnore.contains(U); })) ValuesToIgnore.insert(Op); VecValuesToIgnore.insert(Op); append_range(DeadOps, Op->operands()); } // 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_range(Casts); } // Ignore type-casting instructions we identified during induction // detection. for (const auto &Induction : Legal->getInductionVars()) { const InductionDescriptor &IndDes = Induction.second; VecValuesToIgnore.insert_range(IndDes.getCastInsts()); } } void LoopVectorizationPlanner::plan(ElementCount UserVF, unsigned UserIC) { CM.collectValuesToIgnore(); Config.collectElementTypesForWidening(&CM.ValuesToIgnore); FixedScalableVFPair MaxFactors = CM.computeMaxVF(UserVF, UserIC); if (!MaxFactors) // Cases that should not to be vectorized nor interleaved. return; Config.collectInLoopReductions(); // Cases that may be vectorized may be optimized by unit stride predicates. // TODO: Currently unit stride predicates are added unconditionally, even if // they are not used for the selected VF (e.g. when only interleaving). if (MaxFactors.FixedVF.isVector() || MaxFactors.ScalableVF.isVector()) Legal->collectUnitStridePredicates(); auto VPlan1 = tryToBuildVPlan1(); if (!VPlan1) return; if (!OrigLoop->isInnermost()) { // For outer loops, computeMaxVF returns a single non-scalar VF; build a // plan for that VF only. ElementCount VF = MaxFactors.FixedVF ? MaxFactors.FixedVF : MaxFactors.ScalableVF; buildVPlans(*VPlan1, VF, VF); LLVM_DEBUG(printPlans(dbgs())); return; } // Compute the minimal bitwidths required for integer operations in the loop // for later use by the cost model. Config.computeMinimalBitwidths(); // 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(); } if (CM.foldTailByMasking()) Legal->prepareToFoldTailByMasking(); ElementCount MaxUserVF = UserVF.isScalable() ? MaxFactors.ScalableVF : MaxFactors.FixedVF; if (UserVF) { if (!ElementCount::isKnownLE(UserVF, MaxUserVF)) { reportVectorizationInfo( "UserVF ignored because it may be larger than the maximal safe VF", "InvalidUserVF", ORE, OrigLoop); } else { 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.collectNonVectorizedAndSetWideningDecisions(UserVF); ElementCount EpilogueUserVF = EpilogueVectorizationForceVF; if (EpilogueUserVF.isVector() && ElementCount::isKnownLT(EpilogueUserVF, UserVF)) { CM.collectNonVectorizedAndSetWideningDecisions(EpilogueUserVF); buildVPlans(*VPlan1, EpilogueUserVF, EpilogueUserVF); } buildVPlans(*VPlan1, UserVF, UserVF); if (!VPlans.empty() && VPlans.back()->getSingleVF() == UserVF) { // For scalar VF, skip VPlan cost check as VPlan cost is designed for // vector VFs only. if (UserVF.isScalar() || cost(*VPlans.back(), UserVF, /*RU=*/nullptr).isValid()) { LLVM_DEBUG(dbgs() << "LV: Using user VF " << UserVF << ".\n"); LLVM_DEBUG(printPlans(dbgs())); return; } } VPlans.clear(); reportVectorizationInfo("UserVF ignored because of invalid costs.", "InvalidCost", ORE, OrigLoop); } } // Collect the Vectorization Factor Candidates. SmallVector VFCandidates; for (auto VF = ElementCount::getFixed(1); ElementCount::isKnownLE(VF, MaxFactors.FixedVF); VF *= 2) VFCandidates.push_back(VF); for (auto VF = ElementCount::getScalable(1); ElementCount::isKnownLE(VF, MaxFactors.ScalableVF); VF *= 2) VFCandidates.push_back(VF); for (const auto &VF : VFCandidates) { // Collect Uniform and Scalar instructions after vectorization with VF. CM.collectNonVectorizedAndSetWideningDecisions(VF); } buildVPlans(*VPlan1, ElementCount::getFixed(1), MaxFactors.FixedVF); buildVPlans(*VPlan1, ElementCount::getScalable(1), MaxFactors.ScalableVF); LLVM_DEBUG(printPlans(dbgs())); } InstructionCost VPCostContext::getLegacyCost(Instruction *UI, ElementCount VF) const { InstructionCost Cost = CM.getInstructionCost(UI, VF); if (Cost.isValid() && ForceTargetInstructionCost.getNumOccurrences()) return InstructionCost(ForceTargetInstructionCost); return Cost; } bool VPCostContext::skipCostComputation(Instruction *UI, bool IsVector) const { return CM.ValuesToIgnore.contains(UI) || (IsVector && CM.VecValuesToIgnore.contains(UI)) || SkipCostComputation.contains(UI); } void VPCostContext::invalidateWideningDecision(Instruction *I, ElementCount VF) { CM.setWideningDecision(I, VF, LoopVectorizationCostModel::CM_InvalidatedDecision, 0); } uint64_t VPCostContext::getPredBlockCostDivisor(BasicBlock *BB) const { return CM.getPredBlockCostDivisor(CostKind, BB); } bool VPCostContext::willBeScalarized(Instruction *I, ElementCount VF) const { return CM.isScalarWithPredication(I, VF) || CM.isUniformAfterVectorization(I, VF) || CM.isForcedScalar(I, VF) || (VF.isVector() && CM.isProfitableToScalarize(I, VF)); } bool VPCostContext::isMaskRequired(Instruction *I) const { return CM.isMaskRequired(I); } InstructionCost LoopVectorizationPlanner::precomputeCosts(VPlan &Plan, ElementCount VF, VPCostContext &CostCtx) const { InstructionCost Cost; // Cost modeling for inductions is inaccurate in the legacy cost model // compared to the recipes that are generated. To match here initially during // VPlan cost model bring up directly use the induction costs from the legacy // cost model. Note that we do this as pre-processing; the VPlan may not have // any recipes associated with the original induction increment instruction // and may replace truncates with VPWidenIntOrFpInductionRecipe. We precompute // the cost of induction phis and increments (both that are represented by // recipes and those that are not), to avoid distinguishing between them here, // and skip all recipes that represent induction phis and increments (the // former case) later on, if they exist, to avoid counting them twice. // Similarly we pre-compute the cost of any optimized truncates. // TODO: Switch to more accurate costing based on VPlan. for (const auto &[IV, IndDesc] : Legal->getInductionVars()) { Instruction *IVInc = cast( IV->getIncomingValueForBlock(OrigLoop->getLoopLatch())); SmallVector IVInsts = {IVInc}; for (unsigned I = 0; I != IVInsts.size(); I++) { for (Value *Op : IVInsts[I]->operands()) { auto *OpI = dyn_cast(Op); if (Op == IV || !OpI || !OrigLoop->contains(OpI) || !Op->hasOneUse()) continue; IVInsts.push_back(OpI); } } IVInsts.push_back(IV); for (User *U : IV->users()) { auto *CI = cast(U); if (!CostCtx.CM.isOptimizableIVTruncate(CI, VF)) continue; IVInsts.push_back(CI); } // If the vector loop gets executed exactly once with the given VF, ignore // the costs of comparison and induction instructions, as they'll get // simplified away. // TODO: Remove this code after stepping away from the legacy cost model and // adding code to simplify VPlans before calculating their costs. auto TC = getSmallConstantTripCount(PSE.getSE(), OrigLoop); if (TC == VF && !hasTailFolded(Plan)) addFullyUnrolledInstructionsToIgnore(OrigLoop, Legal->getInductionVars(), CostCtx.SkipCostComputation); for (Instruction *IVInst : IVInsts) { if (CostCtx.skipCostComputation(IVInst, VF.isVector())) continue; InstructionCost InductionCost = CostCtx.getLegacyCost(IVInst, VF); LLVM_DEBUG({ dbgs() << "Cost of " << InductionCost << " for VF " << VF << ": induction instruction " << *IVInst << "\n"; }); Cost += InductionCost; CostCtx.SkipCostComputation.insert(IVInst); } } // Pre-compute the costs for branches except for the backedge, as the number // of replicate regions in a VPlan may not directly match the number of // branches, which would lead to different decisions. // TODO: Compute cost of branches for each replicate region in the VPlan, // which is more accurate than the legacy cost model. for (BasicBlock *BB : OrigLoop->blocks()) { if (CostCtx.skipCostComputation(BB->getTerminator(), VF.isVector())) continue; CostCtx.SkipCostComputation.insert(BB->getTerminator()); if (BB == OrigLoop->getLoopLatch()) continue; auto BranchCost = CostCtx.getLegacyCost(BB->getTerminator(), VF); Cost += BranchCost; } // Don't apply special costs when instruction cost is forced to make sure the // forced cost is used for each recipe. if (ForceTargetInstructionCost.getNumOccurrences()) return Cost; // Pre-compute costs for instructions that are forced-scalar or profitable to // scalarize. For most such instructions, their scalarization costs are // accounted for here using the legacy cost model. However, some opcodes // are excluded from these precomputed scalarization costs and are instead // modeled later by the VPlan cost model (see UseVPlanCostModel below). for (Instruction *ForcedScalar : CM.ForcedScalars[VF]) { if (CostCtx.skipCostComputation(ForcedScalar, VF.isVector())) continue; CostCtx.SkipCostComputation.insert(ForcedScalar); InstructionCost ForcedCost = CostCtx.getLegacyCost(ForcedScalar, VF); LLVM_DEBUG({ dbgs() << "Cost of " << ForcedCost << " for VF " << VF << ": forced scalar " << *ForcedScalar << "\n"; }); Cost += ForcedCost; } auto UseVPlanCostModel = [](Instruction *I) -> bool { switch (I->getOpcode()) { case Instruction::SDiv: case Instruction::UDiv: case Instruction::SRem: case Instruction::URem: return true; default: return false; } }; for (const auto &[Scalarized, ScalarCost] : CM.InstsToScalarize[VF]) { if (UseVPlanCostModel(Scalarized) || CostCtx.skipCostComputation(Scalarized, VF.isVector())) continue; CostCtx.SkipCostComputation.insert(Scalarized); LLVM_DEBUG({ dbgs() << "Cost of " << ScalarCost << " for VF " << VF << ": profitable to scalarize " << *Scalarized << "\n"; }); Cost += ScalarCost; } return Cost; } InstructionCost LoopVectorizationPlanner::cost(VPlan &Plan, ElementCount VF, VPRegisterUsage *RU) const { VPCostContext CostCtx(CM.TTI, *CM.TLI, Plan, CM, Config.CostKind, PSE, OrigLoop); InstructionCost Cost = precomputeCosts(Plan, VF, CostCtx); // Now compute and add the VPlan-based cost. Cost += Plan.cost(VF, CostCtx); // Add the cost of spills due to excess register usage if (RU && Config.shouldConsiderRegPressureForVF(VF)) Cost += RU->spillCost(CM.TTI, Config.CostKind, ForceTargetNumVectorRegs); #ifndef NDEBUG unsigned EstimatedWidth = estimateElementCount(VF, Config.getVScaleForTuning()); LLVM_DEBUG(dbgs() << "Cost for VF " << VF << ": " << Cost << " (Estimated cost per lane: "); if (Cost.isValid()) { APFloat CostPerLane(APFloat::IEEEdouble()); APFloat EstimatedWidthAsAPFloat(APFloat::IEEEdouble()); (void)CostPerLane.convertFromAPInt(APInt(64, (uint64_t)Cost.getValue()), false, APFloat::rmTowardZero); (void)EstimatedWidthAsAPFloat.convertFromAPInt( APInt(64, (uint64_t)EstimatedWidth), false, APFloat::rmTowardZero); (void)CostPerLane.divide(EstimatedWidthAsAPFloat, APFloat::rmTowardZero); SmallString<16> Str; CostPerLane.toString(Str, 3); LLVM_DEBUG(dbgs() << Str); } else /* No point dividing an invalid cost - it will still be invalid */ LLVM_DEBUG(dbgs() << "Invalid"); LLVM_DEBUG(dbgs() << ")\n"); #endif return Cost; } std::pair LoopVectorizationPlanner::computeBestVF() { if (VPlans.empty()) return {VectorizationFactor::Disabled(), nullptr}; // If there is a single VPlan with a single VF, return it directly. VPlan &FirstPlan = *VPlans[0]; ElementCount UserVF = Hints.getWidth(); if (VPlans.size() == 1) { // For outer loops, the plan has a single vector VF determined by the // heuristic. assert((FirstPlan.hasScalarVFOnly() || hasPlanWithVF(UserVF) || FirstPlan.isOuterLoop()) && "must have a single scalar VF, UserVF or an outer loop"); return {VectorizationFactor(FirstPlan.getSingleVF(), 0, 0), &FirstPlan}; } if (hasPlanWithVF(UserVF) && hasForcedEpilogueVF()) { assert(VPlans.size() == 2 && "Must have exactly 2 VPlans built"); assert(VPlans[0]->getSingleVF() == EpilogueVectorizationForceVF && "expected first plan to be for the forced epilogue VF"); assert(VPlans[1]->getSingleVF() == UserVF && "expected second plan to be for the forced UserVF"); return {VectorizationFactor(UserVF, 0, 0), VPlans[1].get()}; } LLVM_DEBUG(dbgs() << "LV: Computing best VF using cost kind: " << (Config.CostKind == TTI::TCK_RecipThroughput ? "Reciprocal Throughput\n" : Config.CostKind == TTI::TCK_Latency ? "Instruction Latency\n" : Config.CostKind == TTI::TCK_CodeSize ? "Code Size\n" : Config.CostKind == TTI::TCK_SizeAndLatency ? "Code Size and Latency\n" : "Unknown\n")); ElementCount ScalarVF = ElementCount::getFixed(1); assert(FirstPlan.hasVF(ScalarVF) && "More than a single plan/VF w/o any plan having scalar VF"); // TODO: Compute scalar cost using VPlan-based cost model. InstructionCost ScalarCost = CM.expectedCost(ScalarVF); LLVM_DEBUG(dbgs() << "LV: Scalar loop costs: " << ScalarCost << ".\n"); VectorizationFactor ScalarFactor(ScalarVF, ScalarCost, ScalarCost); VectorizationFactor BestFactor = ScalarFactor; bool ForceVectorization = Hints.getForce() == LoopVectorizeHints::FK_Enabled; if (ForceVectorization) { // Ignore scalar width, because the user explicitly wants vectorization. // Initialize cost to max so that VF = 2 is, at least, chosen during cost // evaluation. BestFactor.Cost = InstructionCost::getMax(); } VPlan *PlanForBestVF = &FirstPlan; for (auto &P : VPlans) { ArrayRef VFs(P->vectorFactors().begin(), P->vectorFactors().end()); SmallVector RUs; bool ConsiderRegPressure = any_of(VFs, [this](ElementCount VF) { return Config.shouldConsiderRegPressureForVF(VF); }); if (ConsiderRegPressure) RUs = calculateRegisterUsageForPlan(*P, VFs, TTI, CM.ValuesToIgnore); for (unsigned I = 0; I < VFs.size(); I++) { ElementCount VF = VFs[I]; if (VF.isScalar()) continue; if (!ForceVectorization && !willGenerateVectors(*P, VF, TTI)) { LLVM_DEBUG( dbgs() << "LV: Not considering vector loop of width " << VF << " because it will not generate any vector instructions.\n"); continue; } if (Config.OptForSize && !ForceVectorization && hasReplicatorRegion(*P)) { LLVM_DEBUG( dbgs() << "LV: Not considering vector loop of width " << VF << " because it would cause replicated blocks to be generated," << " which isn't allowed when optimizing for size.\n"); continue; } InstructionCost Cost = cost(*P, VF, ConsiderRegPressure ? &RUs[I] : nullptr); VectorizationFactor CurrentFactor(VF, Cost, ScalarCost); if (isMoreProfitable(CurrentFactor, BestFactor, P->hasScalarTail())) { BestFactor = CurrentFactor; PlanForBestVF = P.get(); } // If profitable add it to ProfitableVF list. if (isMoreProfitable(CurrentFactor, ScalarFactor, P->hasScalarTail())) ProfitableVFs.push_back(CurrentFactor); } } VPlan &BestPlan = *PlanForBestVF; assert((BestFactor.Width.isScalar() || BestFactor.ScalarCost > 0) && "when vectorizing, the scalar cost must be computed."); LLVM_DEBUG(dbgs() << "LV: Selecting VF: " << BestFactor.Width << ".\n"); return {BestFactor, &BestPlan}; } DenseMap LoopVectorizationPlanner::executePlan( ElementCount BestVF, unsigned BestUF, VPlan &BestVPlan, InnerLoopVectorizer &ILV, DominatorTree *DT, EpilogueVectorizationKind EpilogueVecKind) { assert(BestVPlan.hasVF(BestVF) && "Trying to execute plan with unsupported VF"); assert(BestVPlan.hasUF(BestUF) && "Trying to execute plan with unsupported UF"); if (BestVPlan.hasEarlyExit()) ++LoopsEarlyExitVectorized; RUN_VPLAN_PASS(VPlanTransforms::replaceWideCanonicalIVWithWideIV, BestVPlan, *PSE.getSE(), CM.TTI, Config.CostKind, BestVF, BestUF, CM.ValuesToIgnore); // TODO: Move to VPlan transform stage once the transition to the VPlan-based // cost model is complete for better cost estimates. RUN_VPLAN_PASS(VPlanTransforms::unrollByUF, BestVPlan, BestUF); RUN_VPLAN_PASS(VPlanTransforms::materializePacksAndUnpacks, BestVPlan); RUN_VPLAN_PASS(VPlanTransforms::materializeBroadcasts, BestVPlan); RUN_VPLAN_PASS(VPlanTransforms::replicateByVF, BestVPlan, BestVF); bool HasBranchWeights = hasBranchWeightMD(*OrigLoop->getLoopLatch()->getTerminator()); if (HasBranchWeights) { std::optional VScale = Config.getVScaleForTuning(); RUN_VPLAN_PASS(VPlanTransforms::addBranchWeightToMiddleTerminator, BestVPlan, BestVF, VScale); } if (CM.maskPartialAliasing()) { assert(BestVPlan.hasTailFolded() && "Expected tail folding to be enabled"); RUN_VPLAN_PASS(VPlanTransforms::materializeAliasMaskCheckBlock, BestVPlan, *CM.Legal->getRuntimePointerChecking()->getDiffChecks(), HasBranchWeights); ++LoopsPartialAliasVectorized; } // Retrieving VectorPH now when it's easier while VPlan still has Regions. VPBasicBlock *VectorPH = cast(BestVPlan.getVectorPreheader()); RUN_VPLAN_PASS(VPlanTransforms::materializeConstantVectorTripCount, BestVPlan, BestVF, BestUF, PSE); RUN_VPLAN_PASS(VPlanTransforms::optimizeForVFAndUF, BestVPlan, BestVF, BestUF, PSE); RUN_VPLAN_PASS(VPlanTransforms::simplifyRecipes, BestVPlan); // Check if scalar epilogue is required, before simplifying constant branches. const bool RequiresScalarEpilogue = requiresScalarEpilogue(BestVPlan, BestVF); if (EpilogueVecKind == EpilogueVectorizationKind::None) RUN_VPLAN_PASS(VPlanTransforms::removeBranchOnConst, BestVPlan, /*OnlyLatches=*/false); if (BestVPlan.getEntry()->getSingleSuccessor() == BestVPlan.getScalarPreheader()) { // TODO: The vector loop would be dead, should not even try to vectorize. ORE->emit([&]() { return OptimizationRemarkAnalysis(DEBUG_TYPE, "VectorizationDead", OrigLoop->getStartLoc(), OrigLoop->getHeader()) << "Created vector loop never executes due to insufficient trip " "count."; }); return DenseMap(); } RUN_VPLAN_PASS(VPlanTransforms::removeDeadRecipes, BestVPlan); RUN_VPLAN_PASS(VPlanTransforms::convertToConcreteRecipes, BestVPlan); // Convert the exit condition to AVLNext == 0 for EVL tail folded loops. RUN_VPLAN_PASS(VPlanTransforms::convertEVLExitCond, BestVPlan); // Regions are dissolved after optimizing for VF and UF, which completely // removes unneeded loop regions first. const bool HasTailFolded = hasTailFolded(BestVPlan); RUN_VPLAN_PASS(VPlanTransforms::dissolveLoopRegions, BestVPlan); // Expand BranchOnTwoConds after dissolution, when latch has direct access to // its successors. RUN_VPLAN_PASS(VPlanTransforms::expandBranchOnTwoConds, BestVPlan); // Convert loops with variable-length stepping after regions are dissolved. RUN_VPLAN_PASS(VPlanTransforms::convertToVariableLengthStep, BestVPlan); // Remove dead back-edges for single-iteration loops with BranchOnCond(true). // Only process loop latches to avoid removing edges from the middle block, // which may be needed for epilogue vectorization. VPlanTransforms::removeBranchOnConst(BestVPlan, /*OnlyLatches=*/true); VPlanTransforms::materializeBackedgeTakenCount(BestVPlan, VectorPH); std::optional MaxRuntimeStep; if (auto MaxVScale = getMaxVScale(*CM.TheFunction, CM.TTI)) MaxRuntimeStep = uint64_t(*MaxVScale) * BestVF.getKnownMinValue() * BestUF; assert((OrigLoop->getUniqueLatchExitBlock() || RequiresScalarEpilogue) && "loops not exiting via the latch without required epilogue?"); VPlanTransforms::materializeVectorTripCount( BestVPlan, VectorPH, HasTailFolded, RequiresScalarEpilogue, &BestVPlan.getVFxUF(), MaxRuntimeStep); VPlanTransforms::materializeFactors(BestVPlan, VectorPH, BestVF); // Limit expansions to VPInstruction to when not vectorizing the epilogue. // Currently this code path still relies on code re-using SCEVs expanded // directly to IR instructions. if (EpilogueVecKind == EpilogueVectorizationKind::None) VPlanTransforms::expandSCEVsToVPInstructions(BestVPlan, *PSE.getSE()); VPlanTransforms::cse(BestVPlan); VPlanTransforms::simplifyRecipes(BestVPlan); // Removing branches and incoming values may expose additional simplification // opportunities. if (VPlanTransforms::removeBranchOnConst(BestVPlan, /*OnlyLatches=*/EpilogueVecKind != EpilogueVectorizationKind::None)) VPlanTransforms::simplifyRecipes(BestVPlan); VPlanTransforms::simplifyKnownEVL(BestVPlan, BestVF, PSE); // 0. Generate SCEV-dependent code in the entry, including TripCount, before // making any changes to the CFG. DenseMap ExpandedSCEVs = VPlanTransforms::expandSCEVs(BestVPlan, *PSE.getSE()); // Perform the actual loop transformation. VPTransformState State(&TTI, BestVF, LI, DT, ILV.AC, ILV.Builder, &BestVPlan, OrigLoop->getParentLoop()); #ifdef EXPENSIVE_CHECKS assert(DT->verify(DominatorTree::VerificationLevel::Fast)); #endif // 1. Set up the skeleton for vectorization, including vector pre-header and // middle block. The vector loop is created during VPlan execution. State.CFG.PrevBB = ILV.createVectorizedLoopSkeleton(); if (VPBasicBlock *ScalarPH = BestVPlan.getScalarPreheader()) replaceVPBBWithIRVPBB(ScalarPH, State.CFG.PrevBB->getSingleSuccessor(), &BestVPlan); VPlanTransforms::removeDeadRecipes(BestVPlan); assert(verifyVPlanIsValid(BestVPlan) && "final VPlan is invalid"); // 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. ScalarEvolution &SE = *PSE.getSE(); for (VPIRBasicBlock *Exit : BestVPlan.getExitBlocks()) { if (!Exit->hasPredecessors()) continue; for (VPRecipeBase &PhiR : Exit->phis()) SE.forgetLcssaPhiWithNewPredecessor(OrigLoop, &cast(PhiR).getIRPhi()); } // Forget the original loop and block dispositions. SE.forgetLoop(OrigLoop); SE.forgetBlockAndLoopDispositions(); ILV.printDebugTracesAtStart(); //===------------------------------------------------===// // // Notice: any optimization or new instruction that go // into the code below should also be implemented in // the cost-model. // //===------------------------------------------------===// // Retrieve loop information before executing the plan, which may remove the // original loop, if it becomes unreachable. MDNode *LID = OrigLoop->getLoopID(); unsigned OrigLoopInvocationWeight = 0; std::optional OrigAverageTripCount = getLoopEstimatedTripCount(OrigLoop, &OrigLoopInvocationWeight); BestVPlan.execute(&State); // 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). VPBasicBlock *HeaderVPBB = vputils::getFirstLoopHeader(BestVPlan, State.VPDT); // 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. bool DisableRuntimeUnroll = !ILV.RTChecks.hasChecks() && !BestVF.isScalar(); updateLoopMetadataAndProfileInfo( HeaderVPBB ? LI->getLoopFor(State.CFG.VPBB2IRBB.lookup(HeaderVPBB)) : nullptr, HeaderVPBB, BestVPlan, EpilogueVecKind == EpilogueVectorizationKind::Epilogue, LID, OrigAverageTripCount, OrigLoopInvocationWeight, estimateElementCount(BestVF * BestUF, Config.getVScaleForTuning()), DisableRuntimeUnroll); // 3. Fix the vectorized code: take care of header phi's, live-outs, // predication, updating analyses. ILV.fixVectorizedLoop(State); ILV.printDebugTracesAtEnd(); return ExpandedSCEVs; } //===--------------------------------------------------------------------===// // EpilogueVectorizerMainLoop //===--------------------------------------------------------------------===// 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"; }); } //===--------------------------------------------------------------------===// // EpilogueVectorizerEpilogueLoop //===--------------------------------------------------------------------===// /// This function creates a new scalar preheader, using the previous one as /// entry block to the epilogue VPlan. The minimum iteration check is being /// represented in VPlan. BasicBlock *EpilogueVectorizerEpilogueLoop::createVectorizedLoopSkeleton() { BasicBlock *NewScalarPH = createScalarPreheader("vec.epilog."); BasicBlock *OriginalScalarPH = NewScalarPH->getSinglePredecessor(); OriginalScalarPH->setName("vec.epilog.iter.check"); VPIRBasicBlock *NewEntry = Plan.createVPIRBasicBlock(OriginalScalarPH); VPBasicBlock *OldEntry = Plan.getEntry(); for (auto &R : make_early_inc_range(*OldEntry)) { // Skip moving VPIRInstructions (including VPIRPhis), which are unmovable by // defining. if (isa(&R)) continue; R.moveBefore(*NewEntry, NewEntry->end()); } VPBlockUtils::reassociateBlocks(OldEntry, NewEntry); Plan.setEntry(NewEntry); // OldEntry is now dead and will be cleaned up when the plan gets destroyed. return OriginalScalarPH; } 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 VPRecipeBuilder::isPredicatedInst(Instruction *I) const { return CM.isPredicatedInst(I); } bool VPRecipeBuilder::prefersVectorizedAddressing() const { return CM.TTI.prefersVectorizedAddressing(); } VPRecipeBase *VPRecipeBuilder::tryToWidenMemory(VPInstruction *VPI, VFRange &Range) { assert((VPI->getOpcode() == Instruction::Load || VPI->getOpcode() == Instruction::Store) && "Must be called with either a load or store"); Instruction *I = VPI->getUnderlyingInstr(); 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; // If a mask is not required, drop it - use unmasked version for safe loads. // TODO: Determine if mask is needed in VPlan. VPValue *Mask = CM.isMaskRequired(I) ? VPI->getMask() : nullptr; // 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 = VPI->getOpcode() == Instruction::Load ? VPI->getOperand(0) : VPI->getOperand(1); if (Consecutive) { Builder.setInsertPoint(VPI); Ptr = Builder.createConsecutiveVectorPointer(Ptr, getLoadStoreType(I), Reverse, VPI->getDebugLoc()); } if (Reverse && Mask) Mask = Builder.createNaryOp(VPInstruction::Reverse, Mask, I->getDebugLoc()); if (VPI->getOpcode() == Instruction::Load) { auto *Load = cast(I); auto *LoadR = new VPWidenLoadRecipe(*Load, Ptr, Mask, Consecutive, *VPI, Load->getDebugLoc()); if (Reverse) { Builder.insert(LoadR); return new VPInstruction(VPInstruction::Reverse, LoadR, {}, {}, LoadR->getDebugLoc()); } return LoadR; } StoreInst *Store = cast(I); VPValue *StoredVal = VPI->getOperand(0); if (Reverse) StoredVal = Builder.createNaryOp(VPInstruction::Reverse, StoredVal, Store->getDebugLoc()); return new VPWidenStoreRecipe(*Store, Ptr, StoredVal, Mask, Consecutive, *VPI, Store->getDebugLoc()); } VPWidenIntOrFpInductionRecipe * VPRecipeBuilder::tryToOptimizeInductionTruncate(VPInstruction *VPI, VFRange &Range) { auto *I = cast(VPI->getUnderlyingInstr()); // 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. if (!LoopVectorizationPlanner::getDecisionAndClampRange( bind_front(&LoopVectorizationCostModel::isOptimizableIVTruncate, CM, I), Range)) return nullptr; auto *WidenIV = cast( VPI->getOperand(0)->getDefiningRecipe()); PHINode *Phi = WidenIV->getPHINode(); VPIRValue *Start = WidenIV->getStartValue(); const InductionDescriptor &IndDesc = WidenIV->getInductionDescriptor(); // Wrap flags from the original induction do not apply to the truncated type, // so do not propagate them. VPIRFlags Flags = VPIRFlags::WrapFlagsTy(false, false); VPValue *Step = vputils::getOrCreateVPValueForSCEVExpr(Plan, IndDesc.getStep()); return new VPWidenIntOrFpInductionRecipe( Phi, Start, Step, &Plan.getVF(), IndDesc, I, Flags, VPI->getDebugLoc()); } bool VPRecipeBuilder::shouldWiden(Instruction *I, VFRange &Range) const { assert((!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); } VPRecipeWithIRFlags *VPRecipeBuilder::tryToWiden(VPInstruction *VPI) { auto *I = VPI->getUnderlyingInstr(); switch (VPI->getOpcode()) { default: return nullptr; case Instruction::SDiv: case Instruction::UDiv: case Instruction::SRem: case Instruction::URem: // If not provably safe, use a masked intrinsic. if (CM.isPredicatedInst(I)) return new VPWidenIntrinsicRecipe( getMaskedDivRemIntrinsic(VPI->getOpcode()), VPI->operands(), I->getType(), {}, {}, VPI->getDebugLoc()); [[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, VPI->operandsWithoutMask(), *VPI, *VPI, VPI->getDebugLoc()); case Instruction::ExtractValue: { SmallVector NewOps(VPI->operandsWithoutMask()); auto *EVI = cast(I); assert(EVI->getNumIndices() == 1 && "Expected one extractvalue index"); unsigned Idx = EVI->getIndices()[0]; NewOps.push_back(Plan.getConstantInt(32, Idx)); return new VPWidenRecipe(*I, NewOps, *VPI, *VPI, VPI->getDebugLoc()); } }; } VPHistogramRecipe *VPRecipeBuilder::widenIfHistogram(VPInstruction *VPI) { if (VPI->getOpcode() != Instruction::Store) return nullptr; auto HistInfo = Legal->getHistogramInfo(cast(VPI->getUnderlyingInstr())); if (!HistInfo) return nullptr; const HistogramInfo *HI = *HistInfo; // FIXME: Support other operations. unsigned Opcode = HI->Update->getOpcode(); assert((Opcode == Instruction::Add || Opcode == Instruction::Sub) && "Histogram update operation must be an Add or Sub"); SmallVector HGramOps; // Bucket address. HGramOps.push_back(VPI->getOperand(1)); // Increment value. HGramOps.push_back(Plan.getOrAddLiveIn(HI->Update->getOperand(1))); // In case of predicated execution (due to tail-folding, or conditional // execution, or both), pass the relevant mask. if (CM.isMaskRequired(HI->Store)) HGramOps.push_back(VPI->getMask()); return new VPHistogramRecipe(Opcode, HGramOps, cast(*VPI), VPI->getDebugLoc()); } bool VPRecipeBuilder::replaceWithFinalIfReductionStore( VPInstruction *VPI, VPBuilder &FinalRedStoresBuilder) { StoreInst *SI; if ((SI = dyn_cast(VPI->getUnderlyingInstr())) && Legal->isInvariantAddressOfReduction(SI->getPointerOperand())) { // Only create recipe for the final invariant store of the reduction. if (Legal->isInvariantStoreOfReduction(SI)) { VPValue *Val = VPI->getOperand(0); VPValue *Addr = VPI->getOperand(1); // We need to store the exiting value of the reduction, so use the blend // if tail folded. if (auto *Blend = VPlanPatternMatch::findUserOf(Val)) Val = Blend; [[maybe_unused]] auto *Rdx = VPlanPatternMatch::findUserOf(Val); assert((!Rdx || Rdx->getBackedgeValue() == Val) && "Store of reduction thats not the backedge value?"); auto *Recipe = new VPReplicateRecipe( SI, {Val, Addr}, true /* IsUniform */, nullptr /*Mask*/, *VPI, *VPI, VPI->getDebugLoc()); FinalRedStoresBuilder.insert(Recipe); } VPI->eraseFromParent(); return true; } return false; } VPSingleDefRecipe *VPRecipeBuilder::handleReplication(VPInstruction *VPI, VFRange &Range) { auto *I = VPI->getUnderlyingInstr(); 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 = VPI->getMask(); } // Note that there is some custom logic to mark some intrinsics as uniform // manually above for scalable vectors, which this assert needs to account for // as well. assert((Range.Start.isScalar() || !IsUniform || !IsPredicated || (Range.Start.isScalable() && isa(I))) && "Should not predicate a uniform recipe"); if (IsUniform) { return VPBuilder::createSingleScalarOp( VPI->getOpcode(), VPI->operandsWithoutMask(), BlockInMask, *VPI, *VPI, VPI->getDebugLoc(), I); } auto *Recipe = new VPReplicateRecipe(I, VPI->operandsWithoutMask(), /*IsSingleScalar=*/false, BlockInMask, *VPI, *VPI, VPI->getDebugLoc()); return Recipe; } VPRecipeBase * VPRecipeBuilder::tryToCreateWidenNonPhiRecipe(VPSingleDefRecipe *R, VFRange &Range) { assert(!R->isPhi() && "phis must be handled earlier"); // First, check for specific widening recipes that deal with optimizing // truncates and memory operations. auto *VPI = cast(R); assert(VPI->getOpcode() != Instruction::Call && "Call should have been handled by makeCallWideningDecisions"); VPRecipeBase *Recipe; if (VPI->getOpcode() == Instruction::Trunc && (Recipe = tryToOptimizeInductionTruncate(VPI, Range))) return Recipe; // All widen recipes below deal only with VF > 1. if (LoopVectorizationPlanner::getDecisionAndClampRange( [&](ElementCount VF) { return VF.isScalar(); }, Range)) return nullptr; Instruction *Instr = R->getUnderlyingInstr(); assert(!is_contained({Instruction::Load, Instruction::Store}, VPI->getOpcode()) && "Should have been handled prior to this!"); if (!shouldWiden(Instr, Range)) return nullptr; if (VPI->getOpcode() == Instruction::GetElementPtr) { auto *GEP = cast(Instr); return new VPWidenGEPRecipe(GEP->getSourceElementType(), VPI->operandsWithoutMask(), *VPI, VPI->getDebugLoc(), GEP); } if (Instruction::isCast(VPI->getOpcode())) { auto *CI = cast(Instr); auto *CastR = cast(VPI); return new VPWidenCastRecipe(CI->getOpcode(), VPI->getOperand(0), CastR->getResultType(), CI, *VPI, *VPI, VPI->getDebugLoc()); } return tryToWiden(VPI); } // To allow RUN_VPLAN_PASS to print the VPlan after VF/UF independent // optimizations. static void printOptimizedVPlan(VPlan &) {} VPlanPtr LoopVectorizationPlanner::tryToBuildVPlan1() { bool IsInnerLoop = OrigLoop->isInnermost(); // Set up loop versioning for inner loops with memory runtime checks. // Outer loops don't have LoopAccessInfo since canVectorizeMemory() is not // called for them. std::optional LVer; if (IsInnerLoop) { const LoopAccessInfo *LAI = Legal->getLAI(); LVer.emplace(*LAI, LAI->getRuntimePointerChecking()->getChecks(), OrigLoop, LI, DT, PSE.getSE()); if (!LAI->getRuntimePointerChecking()->getChecks().empty() && !LAI->getRuntimePointerChecking()->getDiffChecks()) { // Only use noalias metadata when using memory checks guaranteeing no // overlap across all iterations. LVer->prepareNoAliasMetadata(); } } // Create initial base VPlan0, to serve as common starting point for all // candidates built later for specific VF ranges. auto VPlan0 = VPlanTransforms::buildVPlan0(OrigLoop, *LI, Legal->getWidestInductionType(), PSE, LVer ? &*LVer : nullptr); VPDominatorTree VPDT(*VPlan0); if (const LoopAccessInfo *LAI = Legal->getLAI()) RUN_VPLAN_PASS(VPlanTransforms::replaceSymbolicStrides, *VPlan0, PSE, LAI->getSymbolicStrides(), VPDT); RUN_VPLAN_PASS(VPlanTransforms::simplifyRecipes, *VPlan0); RUN_VPLAN_PASS(VPlanTransforms::removeDeadRecipes, *VPlan0); // Create recipes for header phis. For outer loops, reductions, recurrences // and in-loop reductions are empty since legality doesn't detect them. if (!RUN_VPLAN_PASS(VPlanTransforms::createHeaderPhiRecipes, *VPlan0, PSE, *OrigLoop, VPDT, Legal->getInductionVars(), Legal->getReductionVars(), Legal->getFixedOrderRecurrences(), Config.getInLoopReductions(), Hints.allowReordering())) { return nullptr; } if (const LoopAccessInfo *LAI = Legal->getLAI()) RUN_VPLAN_PASS(VPlanTransforms::replaceSymbolicStrides, *VPlan0, PSE, LAI->getSymbolicStrides(), VPDT); // Add surviving induction predicates to PSE and check constraints. bool ForceVectorization = Hints.getForce() == LoopVectorizeHints::FK_Enabled; bool OptForSize = !ForceVectorization && (CM.EpilogueLoweringStatus == CM_EpilogueNotAllowedOptSize || CM.EpilogueLoweringStatus == CM_EpilogueNotAllowedLowTripLoop); unsigned SCEVCheckThreshold = ForceVectorization ? PragmaVectorizeSCEVCheckThreshold : VectorizeSCEVCheckThreshold; if (!RUN_VPLAN_PASS(VPlanTransforms::finalizeSCEVPredicates, *VPlan0, PSE, OptForSize, SCEVCheckThreshold, ORE, OrigLoop)) return nullptr; RUN_VPLAN_PASS(VPlanTransforms::addMiddleCheck, *VPlan0); // If we're vectorizing a loop with an uncountable exit, make sure that the // recipes are safe to handle. // TODO: Remove this once we can properly check the VPlan itself for both // the presence of an uncountable exit and the presence of stores in // the loop inside handleEarlyExits itself. UncountableExitStyle EEStyle = UncountableExitStyle::NoUncountableExit; if (Legal->hasUncountableEarlyExit()) EEStyle = Legal->hasUncountableExitWithSideEffects() ? UncountableExitStyle::MaskedHandleExitInScalarLoop : UncountableExitStyle::ReadOnly; if (!RUN_VPLAN_PASS(VPlanTransforms::handleEarlyExits, *VPlan0, EEStyle, OrigLoop, PSE, *DT, Legal->getAssumptionCache())) { return nullptr; } RUN_VPLAN_PASS(VPlanTransforms::createLoopRegions, *VPlan0, getDebugLocFromInstOrOperands(Legal->getPrimaryInduction())); if (CM.foldTailByMasking()) RUN_VPLAN_PASS(VPlanTransforms::foldTailByMasking, *VPlan0); RUN_VPLAN_PASS(VPlanTransforms::introduceMasksAndLinearize, *VPlan0); return VPlan0; } void LoopVectorizationPlanner::buildVPlans(VPlan &VPlan1, ElementCount MinVF, ElementCount MaxVF) { if (ElementCount::isKnownGT(MinVF, MaxVF)) return; auto MaxVFTimes2 = MaxVF * 2; for (ElementCount VF = MinVF; ElementCount::isKnownLT(VF, MaxVFTimes2);) { VFRange SubRange = {VF, MaxVFTimes2}; auto Plan = tryToBuildVPlan(std::unique_ptr(VPlan1.duplicate()), SubRange); VF = SubRange.End; if (!Plan) continue; // Now optimize the initial VPlan. RUN_VPLAN_PASS(VPlanTransforms::hoistPredicatedLoads, *Plan, PSE, OrigLoop); RUN_VPLAN_PASS(VPlanTransforms::sinkPredicatedStores, *Plan, PSE, OrigLoop); RUN_VPLAN_PASS(VPlanTransforms::truncateToMinimalBitwidths, *Plan, Config.getMinimalBitwidths()); RUN_VPLAN_PASS(VPlanTransforms::optimize, *Plan); // TODO: try to put addExplicitVectorLength close to addActiveLaneMask if (CM.foldTailWithEVL()) { RUN_VPLAN_PASS(VPlanTransforms::addExplicitVectorLength, *Plan, Config.getMaxSafeElements()); RUN_VPLAN_PASS(VPlanTransforms::optimizeEVLMasks, *Plan); } if (auto P = RUN_VPLAN_PASS(VPlanTransforms::narrowInterleaveGroups, *Plan, TTI)) VPlans.push_back(std::move(P)); TailFoldingStyle Style = CM.getTailFoldingStyle(); RUN_VPLAN_PASS(VPlanTransforms::materializeHeaderMask, *Plan, useActiveLaneMask(Style), useActiveLaneMaskForControlFlow(Style)); RUN_VPLAN_PASS_NO_VERIFY(printOptimizedVPlan, *Plan); assert(verifyVPlanIsValid(*Plan) && "VPlan is invalid"); VPlans.push_back(std::move(Plan)); } } VPlanPtr LoopVectorizationPlanner::tryToBuildVPlan(VPlanPtr Plan, VFRange &Range) { // For outer loops, the plan only needs basic recipe conversion and induction // live-out optimization; the full inner-loop recipe building below does not // apply (no widening decisions, interleave groups, reductions, etc.). if (Plan->isOuterLoop()) { for (ElementCount VF : Range) Plan->addVF(VF); if (!RUN_VPLAN_PASS(VPlanTransforms::tryToConvertVPInstructionsToVPRecipes, *Plan, *TLI)) return nullptr; RUN_VPLAN_PASS(VPlanTransforms::optimizeInductionLiveOutUsers, *Plan, PSE); return Plan; } using namespace llvm::VPlanPatternMatch; SmallPtrSet *, 1> InterleaveGroups; // --------------------------------------------------------------------------- // 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. // --------------------------------------------------------------------------- bool RequiresScalarEpilogueCheck = LoopVectorizationPlanner::getDecisionAndClampRange( [this](ElementCount VF) { return !CM.requiresScalarEpilogue(VF.isVector()); }, Range); // Update the branch in the middle block if a scalar epilogue is required. VPBasicBlock *MiddleVPBB = Plan->getMiddleBlock(); if (!RequiresScalarEpilogueCheck && MiddleVPBB->getNumSuccessors() == 2) { auto *BranchOnCond = cast(MiddleVPBB->getTerminator()); assert(MiddleVPBB->getSuccessors()[1] == Plan->getScalarPreheader() && "second successor must be scalar preheader"); BranchOnCond->setOperand(0, Plan->getFalse()); } // 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); TailFoldingStyle Style = CM.getTailFoldingStyle(); // Use NUW for the induction increment if we proved that it won't overflow in // the vector loop or when not folding the tail. In the later case, we know // that the canonical induction increment will not overflow as the vector trip // count is >= increment and a multiple of the increment. VPRegionBlock *LoopRegion = Plan->getVectorLoopRegion(); bool HasNUW = !IVUpdateMayOverflow || Style == TailFoldingStyle::None; if (!HasNUW) { auto *IVInc = LoopRegion->getExitingBasicBlock()->getTerminator()->getOperand(0); assert(match(IVInc, m_VPInstruction( m_Specific(LoopRegion->getCanonicalIV()), m_VPValue())) && "Did not find the canonical IV increment"); LoopRegion->clearCanonicalIVNUW(cast(IVInc)); } // --------------------------------------------------------------------------- // 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 interleave factors must be <= 8 since we // require the (de)interleaveN intrinsics instead of shufflevectors. assert((!Result || !VF.isScalable() || IG->getFactor() <= 8) && "Unsupported interleave factor for scalable vectors"); return Result; }; if (!getDecisionAndClampRange(ApplyIG, Range)) continue; InterleaveGroups.insert(IG); } // --------------------------------------------------------------------------- // Construct wide recipes and apply predication for original scalar // VPInstructions in the loop. // --------------------------------------------------------------------------- VPRecipeBuilder RecipeBuilder(*Plan, Legal, CM, Builder); // Scan the body of the loop in a topological order to visit each basic block // after having visited its predecessor basic blocks. VPBasicBlock *HeaderVPBB = LoopRegion->getEntryBasicBlock(); ReversePostOrderTraversal> RPOT( HeaderVPBB); RUN_VPLAN_PASS(VPlanTransforms::createInLoopReductionRecipes, *Plan, Range.Start); VPCostContext CostCtx(CM.TTI, *CM.TLI, *Plan, CM, Config.CostKind, CM.PSE, OrigLoop); RUN_VPLAN_PASS(VPlanTransforms::makeMemOpWideningDecisions, *Plan, Range, RecipeBuilder, CostCtx); RUN_VPLAN_PASS(VPlanTransforms::makeScalarizationDecisions, *Plan, Range); RUN_VPLAN_PASS(VPlanTransforms::makeCallWideningDecisions, *Plan, Range, RecipeBuilder, CostCtx); // Now process all other blocks and instructions. for (VPBasicBlock *VPBB : VPBlockUtils::blocksOnly(RPOT)) { // Convert input VPInstructions to widened recipes. for (VPRecipeBase &R : make_early_inc_range( make_range(VPBB->getFirstNonPhi(), VPBB->end()))) { // Skip recipes that do not need transforming or have already been // transformed. if (isa(&R) || (isa(R) && Instruction::isCast(cast(R).getOpcode()) && vputils::onlyFirstLaneUsed(R.getVPSingleValue()))) continue; auto *VPI = cast(&R); if (!VPI->getUnderlyingValue()) continue; // TODO: Gradually replace uses of underlying instruction by analyses on // VPlan. Migrate code relying on the underlying instruction from VPlan0 // to construct recipes below to not use the underlying instruction. Instruction *Instr = cast(VPI->getUnderlyingValue()); Builder.setInsertPoint(VPI); VPRecipeBase *Recipe = RecipeBuilder.tryToCreateWidenNonPhiRecipe(VPI, Range); if (!Recipe) Recipe = RecipeBuilder.handleReplication(cast(VPI), Range); if (isa(Recipe) && isa(Instr)) { // Optimized a truncate to VPWidenIntOrFpInductionRecipe. It needs to be // moved to the phi section in the header. Recipe->insertBefore(*HeaderVPBB, HeaderVPBB->getFirstNonPhi()); } else { Builder.insert(Recipe); } if (Recipe->getNumDefinedValues() == 1) { VPI->replaceAllUsesWith(Recipe->getVPSingleValue()); } else { assert(Recipe->getNumDefinedValues() == 0 && "Unexpected multidef recipe"); } R.eraseFromParent(); } } assert(isa(LoopRegion) && !LoopRegion->getEntryBasicBlock()->empty() && "entry block must be set to a VPRegionBlock having a non-empty entry " "VPBasicBlock"); RUN_VPLAN_PASS(VPlanTransforms::adjustFirstOrderRecurrenceMiddleUsers, *Plan, Range); // --------------------------------------------------------------------------- // Transform initial VPlan: Apply previously taken decisions, in order, to // bring the VPlan to its final state. // --------------------------------------------------------------------------- addReductionResultComputation(Plan, RecipeBuilder, Range.Start); // Optimize FindIV reductions to use sentinel-based approach when possible. RUN_VPLAN_PASS(VPlanTransforms::optimizeFindIVReductions, *Plan, PSE, *OrigLoop); RUN_VPLAN_PASS(VPlanTransforms::optimizeInductionLiveOutUsers, *Plan, PSE); // Apply mandatory transformation to handle reductions with multiple in-loop // uses if possible, bail out otherwise. if (!RUN_VPLAN_PASS(VPlanTransforms::handleMultiUseReductions, *Plan, ORE, OrigLoop)) return nullptr; // Apply mandatory transformation to handle FP maxnum/minnum reduction with // NaNs if possible, bail out otherwise. if (!RUN_VPLAN_PASS(VPlanTransforms::handleMaxMinNumReductions, *Plan)) return nullptr; // Create whole-vector selects for find-last recurrences. if (!RUN_VPLAN_PASS(VPlanTransforms::handleFindLastReductions, *Plan)) return nullptr; RUN_VPLAN_PASS(VPlanTransforms::removeBranchOnConst, *Plan, false); // Create partial reduction recipes for scaled reductions and transform // recipes to abstract recipes if it is legal and beneficial and clamp the // range for better cost estimation. // TODO: Enable following transform when the EVL-version of extended-reduction // and mulacc-reduction are implemented. if (!CM.foldTailWithEVL()) { RUN_VPLAN_PASS(VPlanTransforms::createPartialReductions, *Plan, CostCtx, Range); RUN_VPLAN_PASS(VPlanTransforms::convertToAbstractRecipes, *Plan, CostCtx, Range); } // 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. RUN_VPLAN_PASS(VPlanTransforms::createInterleaveGroups, *Plan, InterleaveGroups, CM.isEpilogueAllowed()); // Convert memory recipes to strided access recipes if the strided access is // legal and profitable. RUN_VPLAN_PASS(VPlanTransforms::convertToStridedAccesses, *Plan, PSE, *OrigLoop, CostCtx, Range); // Ensure scalar VF plans only contain VF=1, as required by hasScalarVFOnly. if (Range.Start.isScalar()) Range.End = Range.Start * 2; for (ElementCount VF : Range) Plan->addVF(VF); Plan->setName("Initial VPlan"); RUN_VPLAN_PASS(VPlanTransforms::dropPoisonGeneratingRecipes, *Plan); if (CM.maskPartialAliasing()) RUN_VPLAN_PASS(VPlanTransforms::attachAliasMaskToHeaderMask, *Plan); assert(verifyVPlanIsValid(*Plan) && "VPlan is invalid"); return Plan; } void LoopVectorizationPlanner::addReductionResultComputation( VPlanPtr &Plan, VPRecipeBuilder &RecipeBuilder, ElementCount MinVF) { using namespace VPlanPatternMatch; VPRegionBlock *VectorLoopRegion = Plan->getVectorLoopRegion(); VPBasicBlock *MiddleVPBB = Plan->getMiddleBlock(); VPBasicBlock *LatchVPBB = VectorLoopRegion->getExitingBasicBlock(); Builder.setInsertPoint(&*std::prev(std::prev(LatchVPBB->end()))); VPBasicBlock::iterator IP = MiddleVPBB->getFirstNonPhi(); VPValue *HeaderMask = Plan->getVectorLoopRegion()->getHeaderMask(); for (VPRecipeBase &R : Plan->getVectorLoopRegion()->getEntryBasicBlock()->phis()) { VPReductionPHIRecipe *PhiR = dyn_cast(&R); if (!PhiR) continue; RecurKind RecurrenceKind = PhiR->getRecurrenceKind(); const RecurrenceDescriptor &RdxDesc = Legal->getRecurrenceDescriptor( cast(PhiR->getUnderlyingInstr())); Type *PhiTy = PhiR->getScalarType(); // Convert a VPBlendRecipe backedge to a select. if (auto *Blend = dyn_cast(PhiR->getBackedgeValue())) { if (Blend->getNumIncomingValues() == 2 && Blend->getMask(0) == HeaderMask) { auto *Sel = VPBuilder(Blend).createSelect( Blend->getMask(0), Blend->getIncomingValue(0), Blend->getIncomingValue(1), {}, "", *Blend); Blend->replaceAllUsesWith(Sel); Blend->eraseFromParent(); } } auto *OrigExitingVPV = PhiR->getBackedgeValue(); auto *NewExitingVPV = OrigExitingVPV; // Remove the predicated select if the target doesn't want it. VPValue *V; if (!CM.usePredicatedReductionSelect(RecurrenceKind) && match(PhiR->getBackedgeValue(), m_Select(m_Specific(HeaderMask), m_VPValue(V), m_Specific(PhiR)))) PhiR->setBackedgeValue(V); // 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. VPInstruction *FinalReductionResult; VPBuilder::InsertPointGuard Guard(Builder); Builder.setInsertPoint(MiddleVPBB, IP); // For AnyOf reductions, find the select among PhiR's users and convert // the reduction phi to operate on bools before creating the final // reduction result. if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RecurrenceKind)) { auto *AnyOfSelect = cast( findUserOf(PhiR, m_Select(m_VPValue(), m_VPValue(), m_VPValue()))); VPValue *Start = PhiR->getStartValue(); bool TrueValIsPhi = AnyOfSelect->getOperand(1) == PhiR; // NewVal is the non-phi operand of the select. VPValue *NewVal = TrueValIsPhi ? AnyOfSelect->getOperand(2) : AnyOfSelect->getOperand(1); // Adjust AnyOf reductions; replace the reduction phi for the selected // value with a boolean reduction phi node to check if the condition is // true in any iteration. The final value is selected by the final // ComputeReductionResult. VPValue *Cmp = AnyOfSelect->getOperand(0); // If the compare is checking the reduction PHI node, adjust it to check // the start value. if (VPRecipeBase *CmpR = Cmp->getDefiningRecipe()) CmpR->replaceUsesOfWith(PhiR, PhiR->getStartValue()); Builder.setInsertPoint(AnyOfSelect); // If the true value of the select is the reduction phi, the new value // is selected if the negated condition is true in any iteration. if (TrueValIsPhi) Cmp = Builder.createNot(Cmp); // Build a fresh i1 chain (phi, or, and i1 versions of any blend/select // the exiting value flows through). auto *NewPhiR = PhiR->cloneWithOperands(Plan->getFalse(), Plan->getFalse()); NewPhiR->insertBefore(PhiR); VPValue *NewExiting = Builder.createOr(NewPhiR, Cmp); // The exiting value may flow through a chain of VPBlendRecipes and // select recipes (VPInstruction, VPWidenRecipe or VPReplicateRecipe with // Select opcode) before reaching OrigExitingVPV. Clone each chain link // in topological order so each clone refers to the already-rewritten i1 // operands via Substitutions. DenseMap Substitutions = {{AnyOfSelect, NewExiting}, {PhiR, NewPhiR}}; std::function CloneChain = [&](VPSingleDefRecipe *Old) { if (Substitutions.contains(Old)) return; SmallVector NewOps; for (VPValue *Op : Old->operands()) { if (isa(Op) || match(Op, m_Select(m_VPValue(), m_VPValue(), m_VPValue()))) CloneChain(cast(Op)); NewOps.push_back(Substitutions.lookup_or(Op, Op)); } VPSingleDefRecipe *New; if (auto *B = dyn_cast(Old)) New = B->cloneWithOperands(NewOps); else if (auto *W = dyn_cast(Old)) New = W->cloneWithOperands(NewOps); else if (auto *Rep = dyn_cast(Old)) New = Rep->cloneWithOperands(NewOps); else New = cast(Old)->cloneWithOperands(NewOps); New->insertBefore(Old); Substitutions[Old] = New; }; if (OrigExitingVPV != AnyOfSelect) { CloneChain(cast(OrigExitingVPV)); NewExiting = Substitutions.lookup(OrigExitingVPV); } NewPhiR->setOperand(1, NewExiting); PhiR->replaceAllUsesWith(Plan->getPoison(PhiR->getScalarType())); Builder.setInsertPoint(MiddleVPBB, IP); FinalReductionResult = Builder.createAnyOfReduction(NewExiting, NewVal, Start, ExitDL); } else { // 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. VPValue *ReductionOp = NewExitingVPV; Instruction::CastOps ExtendOpc = Instruction::CastOpsEnd; if (MinVF.isVector() && PhiTy != RdxDesc.getRecurrenceType()) { assert(!PhiR->isInLoop() && "Unexpected truncated inloop reduction!"); assert(!RecurrenceDescriptor::isMinMaxRecurrenceKind(RecurrenceKind) && "Unexpected truncated min-max recurrence!"); Type *RdxTy = RdxDesc.getRecurrenceType(); ExtendOpc = RdxDesc.isSigned() ? Instruction::SExt : Instruction::ZExt; { VPBuilder::InsertPointGuard Guard(Builder); Builder.setInsertPoint( NewExitingVPV->getDefiningRecipe()->getParent(), std::next(NewExitingVPV->getDefiningRecipe()->getIterator())); ReductionOp = Builder.createWidenCast(Instruction::Trunc, NewExitingVPV, RdxTy); VPWidenCastRecipe *Extnd = Builder.createWidenCast(ExtendOpc, ReductionOp, PhiTy); if (PhiR->getOperand(1) == NewExitingVPV) PhiR->setOperand(1, Extnd); } } VPIRFlags Flags(RecurrenceKind, PhiR->isOrdered(), PhiR->isInLoop(), PhiR->getFastMathFlagsOrNone()); FinalReductionResult = Builder.createNaryOp( VPInstruction::ComputeReductionResult, {ReductionOp}, Flags, ExitDL); if (ExtendOpc != Instruction::CastOpsEnd) FinalReductionResult = Builder.createScalarCast( ExtendOpc, FinalReductionResult, PhiTy, {}); } // Update all users outside the vector region. Also replace redundant // extracts. for (auto *U : to_vector(OrigExitingVPV->users())) { auto *Parent = cast(U)->getParent(); if (FinalReductionResult == U || Parent->getParent()) continue; // Skip ComputeReductionResult and FindIV reductions when they are not the // final result. if (match(U, m_VPInstruction()) || (RecurrenceDescriptor::isFindIVRecurrenceKind(RecurrenceKind) && match(U, m_VPInstruction()))) continue; U->replaceUsesOfWith(OrigExitingVPV, FinalReductionResult); // Look through ExtractLastPart. if (match(U, m_ExtractLastPart(m_VPValue()))) U = cast(U)->getSingleUser(); if (match(U, m_CombineOr(m_ExtractLane(m_VPValue(), m_VPValue()), m_ExtractLastLane(m_VPValue())))) cast(U)->replaceAllUsesWith(FinalReductionResult); } RecurKind RK = PhiR->getRecurrenceKind(); if ((!RecurrenceDescriptor::isAnyOfRecurrenceKind(RK) && !RecurrenceDescriptor::isFindIVRecurrenceKind(RK) && !RecurrenceDescriptor::isMinMaxRecurrenceKind(RK) && !RecurrenceDescriptor::isFindLastRecurrenceKind(RK))) { VPBuilder PHBuilder(Plan->getVectorPreheader()); VPValue *Iden = Plan->getOrAddLiveIn( getRecurrenceIdentity(RK, PhiTy, PhiR->getFastMathFlagsOrNone())); auto *ScaleFactorVPV = Plan->getConstantInt(32, 1); VPValue *StartV = PHBuilder.createNaryOp( VPInstruction::ReductionStartVector, {PhiR->getStartValue(), Iden, ScaleFactorVPV}, *PhiR); PhiR->setOperand(0, StartV); } } RUN_VPLAN_PASS(VPlanTransforms::clearReductionWrapFlags, *Plan); } void LoopVectorizationPlanner::attachRuntimeChecks( VPlan &Plan, GeneratedRTChecks &RTChecks, bool HasBranchWeights) const { const auto &[SCEVCheckCond, SCEVCheckBlock] = RTChecks.getSCEVChecks(); if (SCEVCheckBlock && SCEVCheckBlock->hasNPredecessors(0)) { assert((!Config.OptForSize || CM.Hints->getForce() == LoopVectorizeHints::FK_Enabled) && "Cannot SCEV check stride or overflow when optimizing for size"); RUN_VPLAN_PASS(VPlanTransforms::attachCheckBlock, Plan, SCEVCheckCond, SCEVCheckBlock, HasBranchWeights); } const auto &[MemCheckCond, MemCheckBlock] = RTChecks.getMemRuntimeChecks(); if (MemCheckBlock && MemCheckBlock->hasNPredecessors(0)) { // VPlan-native path does not do any analysis for runtime checks // currently. assert((!EnableVPlanNativePath || !Plan.isOuterLoop()) && "Runtime checks are not supported for outer loops yet"); if (Config.OptForSize) { assert( CM.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')."; }); } RUN_VPLAN_PASS(VPlanTransforms::attachCheckBlock, Plan, MemCheckCond, MemCheckBlock, HasBranchWeights); } } bool LoopVectorizationPlanner::requiresScalarEpilogue(VPlan &Plan, ElementCount VF) const { // A scalar epilogue is required, if we unconditionally execute the scalar // loop. Must be called before removeBranchOnConst. VPBasicBlock *MiddleVPBB = Plan.getMiddleBlock(); bool Result = MiddleVPBB->getSingleSuccessor() == Plan.getScalarPreheader(); assert(CM.requiresScalarEpilogue(VF.isVector()) == Result && "CM.requiresScalarEpilogue and the VPlan-based check must agree"); return Result; } bool LoopVectorizationPlanner::hasTailFolded(const VPlan &Plan) const { bool Result = Plan.hasTailFolded(); assert(CM.foldTailByMasking() == Result && "CM.foldTailByMasking and the VPlan-based check must agree"); return Result; } void LoopVectorizationPlanner::addMinimumIterationCheck( VPlan &Plan, ElementCount VF, unsigned UF, ElementCount MinProfitableTripCount) const { const uint32_t *BranchWeights = hasBranchWeightMD(*OrigLoop->getLoopLatch()->getTerminator()) ? &MinItersBypassWeights[0] : nullptr; RUN_VPLAN_PASS(VPlanTransforms::addMinimumIterationCheck, Plan, VF, UF, MinProfitableTripCount, requiresScalarEpilogue(Plan, VF), hasTailFolded(Plan), OrigLoop, BranchWeights, OrigLoop->getLoopPredecessor()->getTerminator()->getDebugLoc(), PSE, Plan.getEntry()); } // Determine how to lower the epilogue, which depends on 1) optimising // for minimum code-size, 2) tail-folding compiler options, 3) loop // hints forcing tail-folding, and 4) a TTI hook that analyses whether the loop // is suitable for tail-folding. // This function determines epilogue lowering for the main vector loop while // epilogue lowering for the tail-folded epilogue path will be handled // separately in getEpilogueTailLowering. static EpilogueLowering getEpilogueLowering(Function *F, Loop *L, LoopVectorizeHints &Hints, bool OptForSize, 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 an epilogue. if (F->hasOptSize() || (OptForSize && Hints.getForce() != LoopVectorizeHints::FK_Enabled)) return CM_EpilogueNotAllowedOptSize; // 2) If set, obey the directives if (TailFoldingPolicy.getNumOccurrences()) { switch (TailFoldingPolicy) { case TailFoldingPolicyTy::None: return CM_EpilogueAllowed; case TailFoldingPolicyTy::PreferFoldTail: return CM_EpilogueNotNeededFoldTail; case TailFoldingPolicyTy::MustFoldTail: return CM_EpilogueNotAllowedFoldTail; }; } // 3) If set, obey the hints switch (Hints.getPredicate()) { case LoopVectorizeHints::FK_Enabled: return CM_EpilogueNotNeededFoldTail; case LoopVectorizeHints::FK_Disabled: return CM_EpilogueAllowed; }; // 4) if the TTI hook indicates this is profitable, request tail-folding. TailFoldingInfo TFI(TLI, &LVL, IAI); if (TTI->preferTailFoldingOverEpilogue(&TFI)) return CM_EpilogueNotNeededFoldTail; return CM_EpilogueAllowed; } /// Determine how to lower the epilogue for the vector epilogue loop. /// Check if there are any conflicts that prevent tail-folding the epilogue. /// \return CM_EpilogueNotNeededFoldTail if epilogue tail-folding is possible, /// otherwise CM_EpilogueAllowed. static EpilogueLowering getEpilogueTailLowering(const LoopVectorizationCostModel &MainCM, const Loop *L, OptimizationRemarkEmitter *ORE) { // Epilogue TF is only enabled when explicitly requested via command line. if (!EpilogueTailFoldingPolicy.getNumOccurrences() || EpilogueTailFoldingPolicy != TailFoldingPolicyTy::PreferFoldTail) return CM_EpilogueAllowed; if (!EnableEpilogueVectorization) { reportVectorizationInfo( "Options conflict, epilogue vectorization is disallowed while " "epilogue tail-folding allowed!\n", "UnsupportedEpilogueTailFoldingPolicy", ORE, L); return CM_EpilogueAllowed; } // If scalar epilogue is explicitly required, we can't apply TF. if (MainCM.requiresScalarEpilogue(/*IsVectorizing*/ true)) { LLVM_DEBUG(dbgs() << "LV: Epilogue tail-folding can't be applied because " "scalar epilogue is required\n" "LV: Fall back to a normal epilogue\n"); return CM_EpilogueAllowed; } // If having epilogue is NOT allowed, then no epilogue to apply TF for. if (!MainCM.isEpilogueAllowed()) { LLVM_DEBUG(dbgs() << "LV: No epilogue to apply tail-folding for.\n" "LV: Fall back to a normal epilogue\n"); return CM_EpilogueAllowed; } // We can apply tail-folding on the vectorized epilogue loop. return CM_EpilogueNotNeededFoldTail; } // 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); } } /// For loops with uncountable early exits, find the cost of doing work when /// exiting the loop early, such as calculating the final exit values of /// variables used outside the loop. /// TODO: This is currently overly pessimistic because the loop may not take /// the early exit, but better to keep this conservative for now. In future, /// it might be possible to relax this by using branch probabilities. static InstructionCost calculateEarlyExitCost(VPCostContext &CostCtx, VPlan &Plan, ElementCount VF) { InstructionCost Cost = 0; for (auto *ExitVPBB : Plan.getExitBlocks()) { for (auto *PredVPBB : ExitVPBB->getPredecessors()) { // If the predecessor is not the middle.block, then it must be the // vector.early.exit block, which may contain work to calculate the exit // values of variables used outside the loop. if (PredVPBB != Plan.getMiddleBlock()) { LLVM_DEBUG(dbgs() << "Calculating cost of work in exit block " << PredVPBB->getName() << ":\n"); Cost += PredVPBB->cost(VF, CostCtx); } } } return Cost; } /// This function determines whether or not it's still profitable to vectorize /// the loop given the extra work we have to do outside of the loop: /// 1. Perform the runtime checks before entering the loop to ensure it's safe /// to vectorize. /// 2. In the case of loops with uncountable early exits, we may have to do /// extra work when exiting the loop early, such as calculating the final /// exit values of variables used outside the loop. /// 3. The middle block. static bool isOutsideLoopWorkProfitable(GeneratedRTChecks &Checks, VectorizationFactor &VF, Loop *L, PredicatedScalarEvolution &PSE, VPCostContext &CostCtx, VPlan &Plan, EpilogueLowering SEL, std::optional VScale) { InstructionCost RtC = Checks.getCost(); if (!RtC.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()) { // TODO: Should we rename VectorizeMemoryCheckThreshold? if (RtC > 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. uint64_t ScalarC = VF.ScalarCost.getValue(); if (ScalarC == 0) return true; InstructionCost TotalCost = RtC; // Add on the cost of any work required in the vector early exit block, if // one exists. TotalCost += calculateEarlyExitCost(CostCtx, Plan, VF.Width); TotalCost += Plan.getMiddleBlock()->cost(VF.Width, CostCtx); // 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 // TotalCost + VecC * (TC / VF) + EpiC // where // * TotalCost is the sum of the costs cost of // - the generated runtime checks, i.e. RtC // - performing any additional work in the vector.early.exit block for // loops with uncountable early exits. // - the middle block, if ExpectedTC <= VF.Width. // * 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: // TotalCost + VecC * (TC / VF) + EpiC < ScalarC * TC // // Now we can compute the minimum required trip count TC as // VF * (TotalCost + EpiC) / (ScalarC * VF - VecC) < 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 = estimateElementCount(VF.Width, VScale); uint64_t Div = ScalarC * IntVF - VF.Cost.getValue(); uint64_t MinTC1 = Div == 0 ? 0 : divideCeil(TotalCost.getValue() * IntVF, Div); // 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 uint64_t MinTC2 = divideCeil(RtC.getValue() * 10, ScalarC); // Now pick the larger minimum. If it is not a multiple of VF and an epilogue // is allowed, choose the next closest multiple of VF. This should partly // compensate for ignoring the epilogue cost. uint64_t MinTC = std::max(MinTC1, MinTC2); if (SEL == CM_EpilogueAllowed) 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(PSE, L)) { if (ElementCount::isKnownLT(*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) {} /// Prepare \p MainPlan for vectorizing the main vector loop during epilogue /// vectorization. static SmallVector preparePlanForMainVectorLoop(VPlan &MainPlan, VPlan &EpiPlan) { using namespace VPlanPatternMatch; // When vectorizing the epilogue, FindFirstIV & FindLastIV reductions can // introduce multiple uses of undef/poison. If the reduction start value may // be undef or poison it needs to be frozen and the frozen start has to be // used when computing the reduction result. We also need to use the frozen // value in the resume phi generated by the main vector loop, as this is also // used to compute the reduction result after the epilogue vector loop. auto AddFreezeForFindLastIVReductions = [](VPlan &Plan, bool UpdateResumePhis) { VPBuilder Builder(Plan.getEntry()); for (VPRecipeBase &R : *Plan.getMiddleBlock()) { auto *VPI = dyn_cast(&R); if (!VPI) continue; VPValue *OrigStart; if (!matchFindIVResult(VPI, m_VPValue(), m_VPValue(OrigStart))) continue; if (isGuaranteedNotToBeUndefOrPoison(OrigStart->getLiveInIRValue())) continue; VPInstruction *Freeze = Builder.createNaryOp(Instruction::Freeze, {OrigStart}, {}, "fr"); VPI->setOperand(2, Freeze); if (UpdateResumePhis) OrigStart->replaceUsesWithIf(Freeze, [Freeze](VPUser &U, unsigned) { return Freeze != &U && isa(&U); }); } }; AddFreezeForFindLastIVReductions(MainPlan, true); AddFreezeForFindLastIVReductions(EpiPlan, false); VPValue *VectorTC = nullptr; auto *Term = MainPlan.getVectorLoopRegion()->getExitingBasicBlock()->getTerminator(); [[maybe_unused]] bool MatchedTC = match(Term, m_BranchOnCount(m_VPValue(), m_VPValue(VectorTC))); assert(MatchedTC && "must match vector trip count"); // If there is a suitable resume value for the canonical induction in the // scalar (which will become vector) epilogue loop, use it and move it to the // beginning of the scalar preheader. Otherwise create it below. VPBasicBlock *MainScalarPH = MainPlan.getScalarPreheader(); auto ResumePhiIter = find_if(MainScalarPH->phis(), [VectorTC](VPRecipeBase &R) { return match(&R, m_VPInstruction(m_Specific(VectorTC), m_ZeroInt())); }); VPPhi *ResumePhi = nullptr; if (ResumePhiIter == MainScalarPH->phis().end()) { assert(MainPlan.getVectorLoopRegion()->getCanonicalIV() && "canonical IV must exist"); Type *Ty = VectorTC->getScalarType(); VPBuilder ScalarPHBuilder(MainScalarPH, MainScalarPH->begin()); ResumePhi = ScalarPHBuilder.createScalarPhi( {VectorTC, MainPlan.getZero(Ty)}, {}, "vec.epilog.resume.val"); } else { ResumePhi = cast(&*ResumePhiIter); ResumePhi->setName("vec.epilog.resume.val"); if (&MainScalarPH->front() != ResumePhi) ResumePhi->moveBefore(*MainScalarPH, MainScalarPH->begin()); } // Create a ResumeForEpilogue for the canonical IV resume and its bypass value // as the first non-phi, to keep them alive for the epilogue. VPBuilder ResumeBuilder(MainScalarPH); ResumeBuilder.createNaryOp(VPInstruction::ResumeForEpilogue, {ResumePhi, ResumePhi->getOperand(1)}); // Create ResumeForEpilogue instructions for the resume phis of the // VPIRPhis and their bypass values in the scalar header of the main plan and // return them so they can be used as resume values when vectorizing the // epilogue. return to_vector( map_range(MainPlan.getScalarHeader()->phis(), [&](VPRecipeBase &R) { assert(isa(R) && "only VPIRPhis expected in the scalar header"); VPValue *MainResumePhi = R.getOperand(0); VPValue *Bypass = MainResumePhi->getDefiningRecipe()->getOperand(1); return ResumeBuilder.createNaryOp(VPInstruction::ResumeForEpilogue, {MainResumePhi, Bypass}); })); } /// Prepare \p Plan for vectorizing the epilogue loop. That is, re-use expanded /// SCEVs from \p ExpandedSCEVs and set resume values for header recipes. Some /// reductions require creating new instructions to compute the resume values. /// They are collected in a vector and returned. They must be moved to the /// preheader of the vector epilogue loop, after created by the execution of \p /// Plan. static SmallVector preparePlanForEpilogueVectorLoop( VPlan &MainPlan, VPlan &Plan, Loop *L, const SCEV2ValueTy &ExpandedSCEVs, EpilogueLoopVectorizationInfo &EPI, LoopVectorizationPlanner &LVP, VFSelectionContext &Config, ScalarEvolution &SE, ArrayRef ResumeValues) { // Build a map from the scalar-header PHI to the ResumeForEpilogue markers // from the main plan. // TODO: Replace the IR PHI key. DenseMap IRPhiToResumeForEpi; for (auto [HeaderPhi, ResumeForEpi] : zip_equal(MainPlan.getScalarHeader()->phis(), ResumeValues)) IRPhiToResumeForEpi[&cast(HeaderPhi).getIRPhi()] = ResumeForEpi; VPRegionBlock *VectorLoop = Plan.getVectorLoopRegion(); VPBasicBlock *Header = VectorLoop->getEntryBasicBlock(); Header->setName("vec.epilog.vector.body"); VPValue *IV = VectorLoop->getCanonicalIV(); // When vectorizing the epilogue loop, the canonical induction needs to start // at the resume value from the main vector loop. Find the resume value // created during execution of the main VPlan. Add this resume value as an // offset to the canonical IV of the epilogue loop. using namespace llvm::PatternMatch; VPInstruction *ResumeForEpilogue = cast(&*MainPlan.getScalarPreheader()->getFirstNonPhi()); Value *EPResumeVal = ResumeForEpilogue->getUnderlyingValue(); if (auto *ResumePhi = dyn_cast(EPResumeVal)) { for (Value *Inc : ResumePhi->incoming_values()) { if (match(Inc, m_SpecificInt(0))) continue; assert(!EPI.VectorTripCount && "Must only have a single non-zero incoming value"); EPI.VectorTripCount = Inc; } // If we didn't find a non-zero vector trip count, all incoming values // must be zero, which also means the vector trip count is zero. if (!EPI.VectorTripCount) { assert(ResumePhi->getNumIncomingValues() > 0 && all_of(ResumePhi->incoming_values(), match_fn(m_SpecificInt(0))) && "all incoming values must be 0"); EPI.VectorTripCount = ResumePhi->getIncomingValue(0); } } else { EPI.VectorTripCount = EPResumeVal; } VPValue *VPV = Plan.getOrAddLiveIn(EPResumeVal); assert(all_of(IV->users(), [](const VPUser *U) { if (isa(U)) return true; unsigned Opc = cast(U)->getOpcode(); return Instruction::isCast(Opc) || Opc == Instruction::Add; }) && "the canonical IV should only be used by its increment or " "ScalarIVSteps when resetting the start value"); VPBuilder Builder(Header, Header->getFirstNonPhi()); VPInstruction *Add = Builder.createAdd(IV, VPV); // Replace all users of the canonical IV and its increment with the offset // version, except for the Add itself and the canonical IV increment. auto *Increment = vputils::findCanonicalIVIncrement(Plan); assert(Increment && "Must have a canonical IV increment at this point"); IV->replaceUsesWithIf(Add, [Add, Increment](VPUser &U, unsigned) { return &U != Add && &U != Increment; }); VPInstruction *OffsetIVInc = VPBuilder::getToInsertAfter(Increment).createAdd(Increment, VPV); Increment->replaceAllUsesWith(OffsetIVInc); OffsetIVInc->setOperand(0, Increment); DenseMap ToFrozen; SmallVector InstsToMove; // Ensure that the start values for all header phi recipes are updated before // vectorizing the epilogue loop. for (VPRecipeBase &R : Header->phis()) { Value *ResumeV = nullptr; // TODO: Move setting of resume values to prepareToExecute. if (auto *ReductionPhi = dyn_cast(&R)) { // Find the reduction result by searching users of the phi or its backedge // value. auto IsReductionResult = [](VPRecipeBase *R) { auto *VPI = dyn_cast(R); return VPI && VPI->getOpcode() == VPInstruction::ComputeReductionResult; }; auto *RdxResult = cast( vputils::findRecipe(ReductionPhi->getBackedgeValue(), IsReductionResult)); assert(RdxResult && "expected to find reduction result"); VPInstruction *ResumeForEpi = IRPhiToResumeForEpi.at( cast(ReductionPhi->getUnderlyingInstr())); ResumeV = ResumeForEpi->getUnderlyingValue(); // Check for FindIV pattern by looking for icmp user of RdxResult. // The pattern is: select(icmp ne RdxResult, Sentinel), RdxResult, Start using namespace VPlanPatternMatch; VPValue *SentinelVPV = nullptr; bool IsFindIV = any_of(RdxResult->users(), [&](VPUser *U) { return match(U, VPlanPatternMatch::m_SpecificICmp( ICmpInst::ICMP_NE, m_Specific(RdxResult), m_VPValue(SentinelVPV))); }); RecurKind RK = ReductionPhi->getRecurrenceKind(); if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK) || IsFindIV) { auto *ResumePhi = cast(ResumeV); VPValue *BypassOp = ResumeForEpi->getOperand(1); assert((isa(BypassOp) || VPlanPatternMatch::match( BypassOp, m_VPInstruction(m_VPValue()))) && "expected live-in or Freeze"); Value *StartV = BypassOp->getUnderlyingValue(); IRBuilder<> Builder(ResumePhi->getParent(), ResumePhi->getParent()->getFirstNonPHIIt()); if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) { // VPReductionPHIRecipes for AnyOf reductions expect a boolean as // start value; compare the final value from the main vector loop // to the start value. ResumeV = Builder.CreateICmpNE(ResumeV, StartV); if (auto *I = dyn_cast(ResumeV)) InstsToMove.push_back(I); } else { assert(SentinelVPV && "expected to find icmp using RdxResult"); if (auto *FreezeI = dyn_cast(StartV)) ToFrozen[FreezeI->getOperand(0)] = StartV; // Adjust resume: select(icmp eq ResumeV, StartV), Sentinel, ResumeV Value *Cmp = Builder.CreateICmpEQ(ResumeV, StartV); if (auto *I = dyn_cast(Cmp)) InstsToMove.push_back(I); ResumeV = Builder.CreateSelect(Cmp, SentinelVPV->getLiveInIRValue(), ResumeV); if (auto *I = dyn_cast(ResumeV)) InstsToMove.push_back(I); } } else { VPValue *StartVal = Plan.getOrAddLiveIn(ResumeV); auto *PhiR = dyn_cast(&R); if (auto *VPI = dyn_cast(PhiR->getStartValue())) { assert(VPI->getOpcode() == VPInstruction::ReductionStartVector && "unexpected start value"); // Partial sub-reductions always start at 0 and account for the // reduction start value in a final subtraction. Update it to use the // resume value from the main vector loop. if (PhiR->getVFScaleFactor() > 1 && RecurrenceDescriptor::isSubRecurrenceKind( PhiR->getRecurrenceKind())) { auto *Sub = cast(RdxResult->getSingleUser()); assert((Sub->getOpcode() == Instruction::Sub || Sub->getOpcode() == Instruction::FSub) && "Unexpected opcode"); assert(isa(Sub->getOperand(0)) && "Expected operand to match the original start value of the " "reduction"); // For integer sub-reductions, verify start value is zero. // For FP sub-reductions, verify start value is negative zero. [[maybe_unused]] auto StartValueIsIdentity = [&] { Value *IdentityValue = getRecurrenceIdentity( PhiR->getRecurrenceKind(), ResumeV->getType(), PhiR->getFastMathFlagsOrNone()); auto *StartValue = dyn_cast(VPI->getOperand(0)); return StartValue && StartValue->getValue() == IdentityValue; }; assert(StartValueIsIdentity() && "Expected start value for partial sub-reduction to be zero " "(or negative zero)"); Sub->setOperand(0, StartVal); } else VPI->setOperand(0, StartVal); continue; } } } else { // Retrieve the induction resume value via ResumeForEpilogue. PHINode *IndPhi = cast(&R)->getPHINode(); ResumeV = IRPhiToResumeForEpi.at(IndPhi)->getUnderlyingValue(); } assert(ResumeV && "Must have a resume value"); VPValue *StartVal = Plan.getOrAddLiveIn(ResumeV); cast(&R)->setStartValue(StartVal); } // For some VPValues in the epilogue plan we must re-use the generated IR // values from the main plan. Replace them with live-in VPValues. // TODO: This is a workaround needed for epilogue vectorization and it // should be removed once induction resume value creation is done // directly in VPlan. for (auto &R : make_early_inc_range(*Plan.getEntry())) { // Re-use frozen values from the main plan for Freeze VPInstructions in the // epilogue plan. This ensures all users use the same frozen value. auto *VPI = dyn_cast(&R); if (VPI && VPI->getOpcode() == Instruction::Freeze) { VPI->replaceAllUsesWith(Plan.getOrAddLiveIn( ToFrozen.lookup(VPI->getOperand(0)->getLiveInIRValue()))); continue; } // 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 auto *ExpandR = dyn_cast(&R); if (!ExpandR) continue; VPValue *ExpandedVal = Plan.getOrAddLiveIn(ExpandedSCEVs.lookup(ExpandR->getSCEV())); ExpandR->replaceAllUsesWith(ExpandedVal); if (Plan.getTripCount() == ExpandR) Plan.resetTripCount(ExpandedVal); ExpandR->eraseFromParent(); } auto VScale = Config.getVScaleForTuning(); unsigned MainLoopStep = estimateElementCount(EPI.MainLoopVF * EPI.MainLoopUF, VScale); unsigned EpilogueLoopStep = estimateElementCount(EPI.EpilogueVF * EPI.EpilogueUF, VScale); RUN_VPLAN_PASS( VPlanTransforms::addMinimumVectorEpilogueIterationCheck, Plan, EPI.VectorTripCount, LVP.requiresScalarEpilogue(Plan, EPI.EpilogueVF), EPI.EpilogueVF, EPI.EpilogueUF, MainLoopStep, EpilogueLoopStep, SE); return InstsToMove; } static void fixScalarResumeValuesFromBypass(BasicBlock *BypassBlock, Loop *L, VPlan &BestEpiPlan, ArrayRef ResumeValues) { // Fix resume values from the additional bypass block. BasicBlock *PH = L->getLoopPreheader(); for (auto *Pred : predecessors(PH)) { for (PHINode &Phi : PH->phis()) { if (Phi.getBasicBlockIndex(Pred) != -1) continue; Phi.addIncoming(Phi.getIncomingValueForBlock(BypassBlock), Pred); } } auto *ScalarPH = cast(BestEpiPlan.getScalarPreheader()); if (ScalarPH->hasPredecessors()) { // Fix resume values for inductions and reductions from the additional // bypass block using the incoming values from the main loop's resume phis. // ResumeValues correspond 1:1 with the scalar loop header phis. for (auto [ResumeV, HeaderPhi] : zip(ResumeValues, BestEpiPlan.getScalarHeader()->phis())) { auto *HeaderPhiR = cast(&HeaderPhi); auto *EpiResumePhi = cast(HeaderPhiR->getIRPhi().getIncomingValueForBlock(PH)); if (EpiResumePhi->getBasicBlockIndex(BypassBlock) == -1) continue; auto *MainResumePhi = cast(ResumeV->getUnderlyingValue()); EpiResumePhi->setIncomingValueForBlock( BypassBlock, MainResumePhi->getIncomingValueForBlock(BypassBlock)); } } } /// Connect the epilogue vector loop generated for \p EpiPlan to the main vector /// loop, after both plans have executed, updating branches from the iteration /// and runtime checks of the main loop, as well as updating various phis. \p /// InstsToMove contains instructions that need to be moved to the preheader of /// the epilogue vector loop. static void connectEpilogueVectorLoop(VPlan &EpiPlan, Loop *L, EpilogueLoopVectorizationInfo &EPI, DominatorTree *DT, GeneratedRTChecks &Checks, ArrayRef InstsToMove, ArrayRef ResumeValues) { BasicBlock *VecEpilogueIterationCountCheck = cast(EpiPlan.getEntry())->getIRBasicBlock(); BasicBlock *VecEpiloguePreHeader = cast(VecEpilogueIterationCountCheck->getTerminator()) ->getSuccessor(1); // 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."); DomTreeUpdater DTU(DT, DomTreeUpdater::UpdateStrategy::Eager); // Helper to redirect an edge from \p BB to \p VecEpilogueIterationCountCheck // to \p NewSucc instead, updating the DomTree. auto RedirectEdge = [&](BasicBlock *BB, BasicBlock *NewSucc) { BB->getTerminator()->replaceUsesOfWith(VecEpilogueIterationCountCheck, NewSucc); DTU.applyUpdates( {{DominatorTree::Delete, BB, VecEpilogueIterationCountCheck}, {DominatorTree::Insert, BB, NewSucc}}); }; RedirectEdge(EPI.MainLoopIterationCountCheck, VecEpiloguePreHeader); BasicBlock *ScalarPH = cast(EpiPlan.getScalarPreheader())->getIRBasicBlock(); RedirectEdge(EPI.EpilogueIterationCountCheck, ScalarPH); // Adjust the terminators of runtime check blocks and phis using them. BasicBlock *SCEVCheckBlock = Checks.getSCEVChecks().second; BasicBlock *MemCheckBlock = Checks.getMemRuntimeChecks().second; if (SCEVCheckBlock) RedirectEdge(SCEVCheckBlock, ScalarPH); if (MemCheckBlock) RedirectEdge(MemCheckBlock, ScalarPH); // 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( llvm::make_pointer_range(VecEpilogueIterationCountCheck->phis())); for (PHINode *Phi : PhisInBlock) { Phi->moveBefore(VecEpiloguePreHeader->getFirstNonPHIIt()); 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; for (BasicBlock *BB : {EPI.EpilogueIterationCountCheck, SCEVCheckBlock, MemCheckBlock}) { if (BB) Phi->removeIncomingValue(BB); } } auto IP = VecEpiloguePreHeader->getFirstNonPHIIt(); for (auto *I : InstsToMove) I->moveBefore(IP); // VecEpilogueIterationCountCheck conditionally skips over the epilogue loop // after executing the main loop. We need to update the resume values of // inductions and reductions during epilogue vectorization. fixScalarResumeValuesFromBypass(VecEpilogueIterationCountCheck, L, EpiPlan, ResumeValues); // Remove dead phis that were moved to the epilogue preheader but are unused // (e.g., resume phis for inductions not widened in the epilogue vector loop). for (PHINode &Phi : make_early_inc_range(VecEpiloguePreHeader->phis())) if (Phi.use_empty()) Phi.eraseFromParent(); } bool LoopVectorizePass::processLoop(Loop *L) { assert((EnableVPlanNativePath || L->isInnermost()) && "VPlan-native path is not enabled. Only process inner loops."); LLVM_DEBUG(dbgs() << "\nLV: Checking a loop in '" << L->getHeader()->getParent()->getName() << "' from " << L->getLocStr() << "\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); // Query this against the original loop and save it here because the profile // of the original loop header may change as the transformation happens. bool OptForSize = llvm::shouldOptimizeForSize( L->getHeader(), PSI, PSI && PSI->hasProfileSummary() ? &GetBFI() : nullptr, PGSOQueryType::IRPass); // 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, /*AllowRuntimeSCEVChecks=*/!OptForSize, AA); if (!LVL.canVectorize(EnableVPlanNativePath)) { LLVM_DEBUG(dbgs() << "LV: Not vectorizing: Cannot prove legality.\n"); Hints.emitRemarkWithHints(); return false; } bool IsInnerLoop = L->isInnermost(); // Outer loops require a computable trip count. if (!IsInnerLoop && isa(PSE.getBackedgeTakenCount())) { LLVM_DEBUG(dbgs() << "LV: cannot compute the outer-loop trip count\n"); return false; } if (LVL.hasUncountableEarlyExit()) { if (!EnableEarlyExitVectorization) { reportVectorizationFailure("Auto-vectorization of loops with uncountable " "early exit is not enabled", "UncountableEarlyExitLoopsDisabled", ORE, L); return false; } if (LVL.hasUncountableExitWithSideEffects() && !EnableEarlyExitVectorizationWithSideEffects) { reportVectorizationFailure("Auto-vectorization of loops with uncountable " "early exit and side effects is not enabled", "UncountableEarlyExitSideEffectLoopsDisabled", ORE, L); return false; } } InterleavedAccessInfo IAI(PSE, L, DT, LI, LVL.getLAI(), OptForSize); bool UseInterleaved = IsInnerLoop && TTI->enableInterleavedAccessVectorization(); // If an override option has been passed in for interleaved accesses, use it. if (EnableInterleavedMemAccesses.getNumOccurrences() > 0) UseInterleaved = IsInnerLoop && EnableInterleavedMemAccesses; // Analyze interleaved memory accesses. if (UseInterleaved) IAI.analyzeInterleaving(useMaskedInterleavedAccesses(*TTI)); if (LVL.hasUncountableEarlyExit()) { BasicBlock *LoopLatch = L->getLoopLatch(); if (IAI.requiresScalarEpilogue() || any_of(LVL.getCountableExitingBlocks(), not_equal_to(LoopLatch))) { reportVectorizationFailure("Auto-vectorization of early exit loops " "requiring a scalar epilogue is unsupported", "UncountableEarlyExitUnsupported", ORE, L); return false; } } // Check the function attributes and profiles to find out if this function // should be optimized for size. EpilogueLowering SEL = getEpilogueLowering(F, L, Hints, OptForSize, 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(PSE, L); if (ExpectedTC && ExpectedTC->isFixed() && ExpectedTC->getFixedValue() < 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 { LLVM_DEBUG(dbgs() << "\n"); // Tail-folded loops are efficient even when the loop // iteration count is low. However, setting the epilogue policy to // `CM_EpilogueNotAllowedLowTripLoop` prevents vectorizing loops // with runtime checks. It's more effective to let // `isOutsideLoopWorkProfitable` determine if vectorization is // beneficial for the loop. if (SEL != CM_EpilogueNotNeededFoldTail) SEL = CM_EpilogueNotAllowedLowTripLoop; } } // 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. VFSelectionContext Config(*TTI, &LVL, L, *F, PSE, DB, ORE, &Hints, OptForSize); LoopVectorizationCostModel CM(SEL, L, PSE, LI, &LVL, *TTI, TLI, AC, ORE, GetBFI, F, &Hints, IAI, Config); // Use the planner for vectorization. LoopVectorizationPlanner LVP(L, LI, DT, TLI, *TTI, &LVL, CM, Config, IAI, PSE, Hints, ORE); EpilogueLowering EpilogueTailLoweringStatus = getEpilogueTailLowering(CM, L, ORE); if (EpilogueTailLoweringStatus == EpilogueLowering::CM_EpilogueNotNeededFoldTail) { // TODO: Apply tail-folding on the vectorized epilogue loop. LLVM_DEBUG(dbgs() << "LV: epilogue tail-folding is not supported yet\n"); reportVectorizationInfo( "The epilogue-tail-folding policy prefer-fold-tail is not supported " "yet, fall back to a normal epilogue", "UnsupportedEpilogueTailFoldingPolicy", ORE, L); } // Get user vectorization factor and interleave count. ElementCount UserVF = Hints.getWidth(); unsigned UserIC = Hints.getInterleave(); // Outer loops don't have LoopAccessInfo, so skip the safety check and reset // UserIC (interleaving is not supported for outer loops). if (!IsInnerLoop) UserIC = 0; else if (UserIC > 1 && !LVL.isSafeForAnyVectorWidth()) UserIC = 1; // Plan how to best vectorize. LVP.plan(UserVF, UserIC); auto [VF, BestPlanPtr] = LVP.computeBestVF(); unsigned IC = 1; // For VPlan build stress testing of outer loops, bail after plan // construction. if (!IsInnerLoop && VPlanBuildOuterloopStressTest) return false; if (IsInnerLoop && ORE->allowExtraAnalysis(LV_NAME)) LVP.emitInvalidCostRemarks(ORE); assert((IsInnerLoop || !CM.maskPartialAliasing()) && "Did not expect to alias-mask outer loop"); GeneratedRTChecks Checks(PSE, DT, LI, TTI, Config.CostKind, CM.maskPartialAliasing()); if (IsInnerLoop && LVP.hasPlanWithVF(VF.Width)) { // Select the interleave count. IC = LVP.selectInterleaveCount(*BestPlanPtr, 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, *ORE); // Bail out early if either the SCEV or memory runtime checks are known to // fail. In that case, the vector loop would never execute. using namespace llvm::PatternMatch; if (Checks.getSCEVChecks().first && match(Checks.getSCEVChecks().first, m_One())) return false; if (Checks.getMemRuntimeChecks().first && match(Checks.getMemRuntimeChecks().first, m_One())) return false; } // Check if it is profitable to vectorize with runtime checks. bool ForceVectorization = Hints.getForce() == LoopVectorizeHints::FK_Enabled; VPCostContext CostCtx(CM.TTI, *CM.TLI, *BestPlanPtr, CM, Config.CostKind, CM.PSE, L); if (!ForceVectorization && !isOutsideLoopWorkProfitable(Checks, VF, L, PSE, CostCtx, *BestPlanPtr, SEL, Config.getVScaleForTuning())) { 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 = { "VectorizationNotBeneficial", "the cost-model indicates that vectorization is not beneficial"}; VectorizeLoop = false; } if (UserIC == 1 && Hints.getInterleave() > 1) { assert(!LVL.isSafeForAnyVectorWidth() && "UserIC should only be ignored due to unsafe dependencies"); LLVM_DEBUG(dbgs() << "LV: Ignoring user-specified interleave count.\n"); IntDiagMsg = {"InterleavingUnsafe", "Ignoring user-specified interleave count due to possibly " "unsafe dependencies in the loop."}; InterleaveLoop = false; } else if (!LVP.hasPlanWithVF(VF.Width) && 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 = {"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 = { "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.\n"); IntDiagMsg = {"InterleavingBeneficialButDisabled", "the cost-model indicates that interleaving is beneficial " "but is explicitly disabled or interleave count is set to 1"}; InterleaveLoop = false; } // If there is a histogram in the loop, do not just interleave without // vectorizing. The order of operations will be incorrect without the // histogram intrinsics, which are only used for recipes with VF > 1. if (!VectorizeLoop && InterleaveLoop && LVL.hasHistograms()) { LLVM_DEBUG(dbgs() << "LV: Not interleaving without vectorization due " << "to histogram operations.\n"); IntDiagMsg = { "HistogramPreventsScalarInterleaving", "Unable to interleave without vectorization due to constraints on " "the order of histogram operations"}; InterleaveLoop = false; } // Override IC if user provided an interleave count. IC = UserIC > 0 ? UserIC : IC; if (CM.maskPartialAliasing()) { LLVM_DEBUG( dbgs() << "LV: Not interleaving due to partial aliasing vectorization.\n"); IntDiagMsg = { "PartialAliasingVectorization", "Unable to interleave due to partial aliasing vectorization."}; InterleaveLoop = false; IC = 1; } // FIXME: Enable interleaving for EE-with-side-effects. if (InterleaveLoop && LVL.hasUncountableExitWithSideEffects()) { LLVM_DEBUG(dbgs() << "LV: Not interleaving due to EE with side effects.\n"); IntDiagMsg = {"EEWithSideEffectsPreventsInterleaving", "Unable to interleave due to early exit with side effects."}; InterleaveLoop = false; IC = 1; } // Emit diagnostic messages, if any. if (!VectorizeLoop && !InterleaveLoop) { // Do not vectorize or interleaving the loop. ORE->emit([&]() { return OptimizationRemarkMissed(LV_NAME, VecDiagMsg.first, L->getStartLoc(), L->getHeader()) << VecDiagMsg.second; }); ORE->emit([&]() { return OptimizationRemarkMissed(LV_NAME, IntDiagMsg.first, L->getStartLoc(), L->getHeader()) << IntDiagMsg.second; }); return false; } if (!VectorizeLoop && InterleaveLoop) { LLVM_DEBUG(dbgs() << "LV: Interleave Count is " << IC << '\n'); ORE->emit([&]() { return OptimizationRemarkAnalysis(LV_NAME, VecDiagMsg.first, L->getStartLoc(), L->getHeader()) << VecDiagMsg.second; }); } else if (VectorizeLoop && !InterleaveLoop) { LLVM_DEBUG(dbgs() << "LV: Found a vectorizable loop (" << VF.Width << ") in " << L->getLocStr() << '\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 " << L->getLocStr() << '\n'); LLVM_DEBUG(dbgs() << "LV: Interleave Count is " << IC << '\n'); } // Report the vectorization decision. if (VF.Width.isScalar()) { using namespace ore; assert(IC > 1); ORE->emit([&]() { return OptimizationRemark(LV_NAME, "Interleaved", L->getStartLoc(), L->getHeader()) << "interleaved loop (interleaved count: " << NV("InterleaveCount", IC) << ")"; }); } else { // Report the vectorization decision. reportVectorization(ORE, L, VF.Width, IC); } if (ORE->allowExtraAnalysis(LV_NAME)) checkMixedPrecision(L, ORE); // If we decided that it is *legal* to interleave or vectorize the loop, then // do it. VPlan &BestPlan = *BestPlanPtr; // Consider vectorizing the epilogue too if it's profitable. std::unique_ptr EpiPlan = LVP.selectBestEpiloguePlan(BestPlan, VF.Width, IC); bool HasBranchWeights = hasBranchWeightMD(*L->getLoopLatch()->getTerminator()); if (EpiPlan) { VPlan &BestEpiPlan = *EpiPlan; VPlan &BestMainPlan = BestPlan; ElementCount EpilogueVF = BestEpiPlan.getSingleVF(); // 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. BestEpiPlan.getMiddleBlock()->setName("vec.epilog.middle.block"); BestEpiPlan.getVectorPreheader()->setName("vec.epilog.ph"); SmallVector ResumeValues = preparePlanForMainVectorLoop(BestMainPlan, BestEpiPlan); EpilogueLoopVectorizationInfo EPI(VF.Width, IC, EpilogueVF, 1, BestEpiPlan); // Add minimum iteration check for the epilogue plan, followed by runtime // checks for the main plan. LVP.addMinimumIterationCheck(BestMainPlan, EPI.EpilogueVF, EPI.EpilogueUF, ElementCount::getFixed(0)); LVP.attachRuntimeChecks(BestMainPlan, Checks, HasBranchWeights); RUN_VPLAN_PASS(VPlanTransforms::addIterationCountCheckBlock, BestMainPlan, EPI.MainLoopVF, EPI.MainLoopUF, LVP.requiresScalarEpilogue(BestMainPlan, EPI.MainLoopVF), L, HasBranchWeights ? MinItersBypassWeights : nullptr, L->getLoopPredecessor()->getTerminator()->getDebugLoc(), PSE); EpilogueVectorizerMainLoop MainILV(L, PSE, LI, DT, TTI, AC, EPI, &CM, Checks, BestMainPlan); auto ExpandedSCEVs = LVP.executePlan( EPI.MainLoopVF, EPI.MainLoopUF, BestMainPlan, MainILV, DT, LoopVectorizationPlanner::EpilogueVectorizationKind::MainLoop); ++LoopsVectorized; // Derive EPI fields from VPlan-generated IR. BasicBlock *EntryBB = cast(BestMainPlan.getEntry())->getIRBasicBlock(); EntryBB->setName("iter.check"); EPI.EpilogueIterationCountCheck = EntryBB; // The check chain is: Entry -> [SCEV] -> [Mem] -> MainCheck -> VecPH. // MainCheck is the non-bypass successor of the last runtime check block // (or Entry if there are no runtime checks). BasicBlock *LastCheck = EntryBB; if (BasicBlock *MemBB = Checks.getMemRuntimeChecks().second) LastCheck = MemBB; else if (BasicBlock *SCEVBB = Checks.getSCEVChecks().second) LastCheck = SCEVBB; BasicBlock *ScalarPH = L->getLoopPreheader(); auto *BI = cast(LastCheck->getTerminator()); EPI.MainLoopIterationCountCheck = BI->getSuccessor(BI->getSuccessor(0) == ScalarPH); // Second pass vectorizes the epilogue and adjusts the control flow // edges from the first pass. EpilogueVectorizerEpilogueLoop EpilogILV(L, PSE, LI, DT, TTI, AC, EPI, &CM, Checks, BestEpiPlan); SmallVector InstsToMove = preparePlanForEpilogueVectorLoop( BestMainPlan, BestEpiPlan, L, ExpandedSCEVs, EPI, LVP, Config, *PSE.getSE(), ResumeValues); LVP.attachRuntimeChecks(BestEpiPlan, Checks, HasBranchWeights); LVP.executePlan( EPI.EpilogueVF, EPI.EpilogueUF, BestEpiPlan, EpilogILV, DT, LoopVectorizationPlanner::EpilogueVectorizationKind::Epilogue); connectEpilogueVectorLoop(BestEpiPlan, L, EPI, DT, Checks, InstsToMove, ResumeValues); ++LoopsEpilogueVectorized; } else { InnerLoopVectorizer LB(L, PSE, LI, DT, TTI, AC, VF.Width, IC, &CM, Checks, BestPlan); LVP.addMinimumIterationCheck(BestPlan, VF.Width, IC, VF.MinProfitableTripCount); LVP.attachRuntimeChecks(BestPlan, Checks, HasBranchWeights); if (!IsInnerLoop) LLVM_DEBUG(dbgs() << "Vectorizing outer loop in \"" << F->getName() << "\"\n"); LVP.executePlan(VF.Width, IC, BestPlan, LB, DT); ++LoopsVectorized; } assert(DT->verify(DominatorTree::VerificationLevel::Fast) && "DT not preserved correctly"); assert(!verifyFunction(*F, &dbgs())); return true; } LoopVectorizeResult LoopVectorizePass::runImpl(Function &F) { // 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), false) < 2 || TTI->getMaxInterleaveFactor(ElementCount::getFixed(1), true) < 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) { LI = &AM.getResult(F); // There are no loops in the function. Return before computing other // expensive analyses. if (LI->empty()) return PreservedAnalyses::all(); SE = &AM.getResult(F); TTI = &AM.getResult(F); DT = &AM.getResult(F); TLI = &AM.getResult(F); AC = &AM.getResult(F); DB = &AM.getResult(F); ORE = &AM.getResult(F); LAIs = &AM.getResult(F); AA = &AM.getResult(F); auto &MAMProxy = AM.getResult(F); PSI = MAMProxy.getCachedResult(*F.getParent()); GetBFI = [&AM, &F]() -> BlockFrequencyInfo & { return AM.getResult(F); }; LoopVectorizeResult Result = runImpl(F); if (!Result.MadeAnyChange) return PreservedAnalyses::all(); PreservedAnalyses PA; if (isAssignmentTrackingEnabled(*F.getParent())) { for (auto &BB : F) RemoveRedundantDbgInstrs(&BB); } PA.preserve(); 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 << '>'; }