diff options
Diffstat (limited to 'lib/Transforms/Vectorize/LoopVectorize.cpp')
| -rw-r--r-- | lib/Transforms/Vectorize/LoopVectorize.cpp | 1941 |
1 files changed, 1941 insertions, 0 deletions
diff --git a/lib/Transforms/Vectorize/LoopVectorize.cpp b/lib/Transforms/Vectorize/LoopVectorize.cpp new file mode 100644 index 0000000..a7ef248 --- /dev/null +++ b/lib/Transforms/Vectorize/LoopVectorize.cpp @@ -0,0 +1,1941 @@ +//===- LoopVectorize.cpp - A Loop Vectorizer ------------------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This is the LLVM loop vectorizer. This pass modifies 'vectorizable' loops +// and generates target-independent LLVM-IR. Legalization of the IR is done +// in the codegen. However, the vectorizes uses (will use) the codegen +// interfaces to generate IR that is likely to result in an optimal binary. +// +// The loop vectorizer combines consecutive loop iteration 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. SingleBlockLoopVectorizer - 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. +//===----------------------------------------------------------------------===// +// +// 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. +// +// Other ideas/concepts are from: +// A. Zaks and D. Nuzman. Autovectorization in GCC-two years later. +// +//===----------------------------------------------------------------------===// +#define LV_NAME "loop-vectorize" +#define DEBUG_TYPE LV_NAME +#include "llvm/Constants.h" +#include "llvm/DerivedTypes.h" +#include "llvm/Instructions.h" +#include "llvm/LLVMContext.h" +#include "llvm/Pass.h" +#include "llvm/Analysis/LoopPass.h" +#include "llvm/Value.h" +#include "llvm/Function.h" +#include "llvm/Analysis/Verifier.h" +#include "llvm/Module.h" +#include "llvm/Type.h" +#include "llvm/ADT/SmallVector.h" +#include "llvm/ADT/StringExtras.h" +#include "llvm/Analysis/AliasAnalysis.h" +#include "llvm/Analysis/AliasSetTracker.h" +#include "llvm/Analysis/ScalarEvolution.h" +#include "llvm/Analysis/Dominators.h" +#include "llvm/Analysis/ScalarEvolutionExpressions.h" +#include "llvm/Analysis/ScalarEvolutionExpander.h" +#include "llvm/Analysis/LoopInfo.h" +#include "llvm/Analysis/ValueTracking.h" +#include "llvm/Transforms/Scalar.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/TargetTransformInfo.h" +#include "llvm/Support/CommandLine.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/DataLayout.h" +#include "llvm/Transforms/Utils/Local.h" +#include <algorithm> +using namespace llvm; + +static cl::opt<unsigned> +VectorizationFactor("force-vector-width", cl::init(0), cl::Hidden, + cl::desc("Set the default vectorization width. Zero is autoselect.")); + +/// We don't vectorize loops with a known constant trip count below this number. +const unsigned TinyTripCountThreshold = 16; + +/// When performing a runtime memory check, do not check more than this +/// number of pointers. Notice that the check is quadratic! +const unsigned RuntimeMemoryCheckThreshold = 2; + +namespace { + +// Forward declarations. +class LoopVectorizationLegality; +class LoopVectorizationCostModel; + +/// SingleBlockLoopVectorizer 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. +/// SingleBlockLoopVectorizer does not perform any vectorization-legality +/// checks, and relies on the caller to check for the different legality +/// aspects. The SingleBlockLoopVectorizer relies on the +/// LoopVectorizationLegality class to provide information about the induction +/// and reduction variables that were found to a given vectorization factor. +class SingleBlockLoopVectorizer { +public: + /// Ctor. + SingleBlockLoopVectorizer(Loop *Orig, ScalarEvolution *Se, LoopInfo *Li, + DominatorTree *dt, LPPassManager *Lpm, + unsigned VecWidth): + OrigLoop(Orig), SE(Se), LI(Li), DT(dt), LPM(Lpm), VF(VecWidth), + Builder(Se->getContext()), Induction(0), OldInduction(0) { } + + // Perform the actual loop widening (vectorization). + void vectorize(LoopVectorizationLegality *Legal) { + ///Create a new empty loop. Unlink the old loop and connect the new one. + createEmptyLoop(Legal); + /// Widen each instruction in the old loop to a new one in the new loop. + /// Use the Legality module to find the induction and reduction variables. + vectorizeLoop(Legal); + // Register the new loop and update the analysis passes. + updateAnalysis(); + } + +private: + /// Create an empty loop, based on the loop ranges of the old loop. + void createEmptyLoop(LoopVectorizationLegality *Legal); + /// Copy and widen the instructions from the old loop. + void vectorizeLoop(LoopVectorizationLegality *Legal); + /// Insert the new loop to the loop hierarchy and pass manager + /// and update the analysis passes. + void updateAnalysis(); + + /// This instruction is un-vectorizable. Implement it as a sequence + /// of scalars. + void scalarizeInstruction(Instruction *Instr); + + /// Create a broadcast instruction. This method generates a broadcast + /// instruction (shuffle) for loop invariant values and for the induction + /// value. If this is the induction variable then we extend it to N, N+1, ... + /// this is needed because each iteration in the loop corresponds to a SIMD + /// element. + Value *getBroadcastInstrs(Value *V); + + /// This is a helper function used by getBroadcastInstrs. It adds 0, 1, 2 .. + /// for each element in the vector. Starting from zero. + Value *getConsecutiveVector(Value* Val); + + /// When we go over instructions in the basic block we rely on previous + /// values within the current basic block or on loop invariant values. + /// When we widen (vectorize) values we place them in the map. If the values + /// are not within the map, they have to be loop invariant, so we simply + /// broadcast them into a vector. + Value *getVectorValue(Value *V); + + /// Get a uniform vector of constant integers. We use this to get + /// vectors of ones and zeros for the reduction code. + Constant* getUniformVector(unsigned Val, Type* ScalarTy); + + typedef DenseMap<Value*, Value*> ValueMap; + + /// The original loop. + Loop *OrigLoop; + // Scev analysis to use. + ScalarEvolution *SE; + // Loop Info. + LoopInfo *LI; + // Dominator Tree. + DominatorTree *DT; + // Loop Pass Manager; + LPPassManager *LPM; + // The vectorization factor to use. + unsigned VF; + + // The builder that we use + IRBuilder<> Builder; + + // --- Vectorization state --- + + /// The vector-loop preheader. + BasicBlock *LoopVectorPreHeader; + /// The scalar-loop preheader. + BasicBlock *LoopScalarPreHeader; + /// Middle Block between the vector and the scalar. + BasicBlock *LoopMiddleBlock; + ///The ExitBlock of the scalar loop. + BasicBlock *LoopExitBlock; + ///The vector loop body. + BasicBlock *LoopVectorBody; + ///The scalar loop body. + BasicBlock *LoopScalarBody; + ///The first bypass block. + BasicBlock *LoopBypassBlock; + + /// The new Induction variable which was added to the new block. + PHINode *Induction; + /// The induction variable of the old basic block. + PHINode *OldInduction; + // Maps scalars to widened vectors. + ValueMap WidenMap; +}; + +/// LoopVectorizationLegality checks if it is legal to vectorize a loop, and +/// to what vectorization factor. +/// This class does not look at the profitability of vectorization, only the +/// legality. This class has two main kinds of checks: +/// * Memory checks - The code in canVectorizeMemory checks if vectorization +/// will change the order of memory accesses in a way that will change the +/// correctness of the program. +/// * Scalars checks - The code in canVectorizeBlock checks for a number +/// of different conditions, such as the availability of a single induction +/// variable, that all types are supported and vectorize-able, etc. +/// This code reflects the capabilities of SingleBlockLoopVectorizer. +/// This class is also used by SingleBlockLoopVectorizer for identifying +/// induction variable and the different reduction variables. +class LoopVectorizationLegality { +public: + LoopVectorizationLegality(Loop *Lp, ScalarEvolution *Se, DataLayout *Dl): + TheLoop(Lp), SE(Se), DL(Dl), Induction(0) { } + + /// This represents the kinds of reductions that we support. + enum ReductionKind { + NoReduction, /// Not a reduction. + IntegerAdd, /// Sum of numbers. + IntegerMult, /// Product of numbers. + IntegerOr, /// Bitwise or logical OR of numbers. + IntegerAnd, /// Bitwise or logical AND of numbers. + IntegerXor /// Bitwise or logical XOR of numbers. + }; + + /// This POD struct holds information about reduction variables. + struct ReductionDescriptor { + // Default C'tor + ReductionDescriptor(): + StartValue(0), LoopExitInstr(0), Kind(NoReduction) {} + + // C'tor. + ReductionDescriptor(Value *Start, Instruction *Exit, ReductionKind K): + StartValue(Start), LoopExitInstr(Exit), Kind(K) {} + + // The starting value of the reduction. + // It does not have to be zero! + Value *StartValue; + // The instruction who's value is used outside the loop. + Instruction *LoopExitInstr; + // The kind of the reduction. + ReductionKind Kind; + }; + + // This POD struct holds information about the memory runtime legality + // check that a group of pointers do not overlap. + struct RuntimePointerCheck { + /// This flag indicates if we need to add the runtime check. + bool Need; + /// Holds the pointers that we need to check. + SmallVector<Value*, 2> Pointers; + }; + + /// ReductionList contains the reduction descriptors for all + /// of the reductions that were found in the loop. + typedef DenseMap<PHINode*, ReductionDescriptor> ReductionList; + + /// Returns true if it is legal to vectorize this loop. + /// This does not mean that it is profitable to vectorize this + /// loop, only that it is legal to do so. + bool canVectorize(); + + /// Returns the Induction variable. + PHINode *getInduction() {return Induction;} + + /// Returns the reduction variables found in the loop. + ReductionList *getReductionVars() { return &Reductions; } + + /// Check if the pointer returned by this GEP is consecutive + /// when the index is vectorized. This happens when the last + /// index of the GEP is consecutive, like the induction variable. + /// This check allows us to vectorize A[idx] into a wide load/store. + bool isConsecutiveGep(Value *Ptr); + + /// Returns true if the value V is uniform within the loop. + bool isUniform(Value *V); + + /// Returns true if this instruction will remain scalar after vectorization. + bool isUniformAfterVectorization(Instruction* I) {return Uniforms.count(I);} + + /// Returns the information that we collected about runtime memory check. + RuntimePointerCheck *getRuntimePointerCheck() {return &PtrRtCheck; } +private: + /// Check if a single basic block loop is vectorizable. + /// At this point we know that this is a loop with a constant trip count + /// and we only need to check individual instructions. + bool canVectorizeBlock(BasicBlock &BB); + + /// When we vectorize loops we may change the order in which + /// we read and write from memory. This method checks if it is + /// legal to vectorize the code, considering only memory constrains. + /// Returns true if BB is vectorizable + bool canVectorizeMemory(BasicBlock &BB); + + /// Returns True, if 'Phi' is the kind of reduction variable for type + /// 'Kind'. If this is a reduction variable, it adds it to ReductionList. + bool AddReductionVar(PHINode *Phi, ReductionKind Kind); + /// Returns true if the instruction I can be a reduction variable of type + /// 'Kind'. + bool isReductionInstr(Instruction *I, ReductionKind Kind); + /// Returns True, if 'Phi' is an induction variable. + bool isInductionVariable(PHINode *Phi); + /// Return true if can compute the address bounds of Ptr within the loop. + bool hasComputableBounds(Value *Ptr); + + /// The loop that we evaluate. + Loop *TheLoop; + /// Scev analysis. + ScalarEvolution *SE; + /// DataLayout analysis. + DataLayout *DL; + + // --- vectorization state --- // + + /// Holds the induction variable. + PHINode *Induction; + /// Holds the reduction variables. + ReductionList Reductions; + /// Allowed outside users. This holds the reduction + /// vars which can be accessed from outside the loop. + SmallPtrSet<Value*, 4> AllowedExit; + /// This set holds the variables which are known to be uniform after + /// vectorization. + SmallPtrSet<Instruction*, 4> Uniforms; + /// We need to check that all of the pointers in this list are disjoint + /// at runtime. + RuntimePointerCheck PtrRtCheck; +}; + +/// 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 VectorTargetTransformInfo to query the different backends +/// for the cost of different operations. +class LoopVectorizationCostModel { +public: + /// C'tor. + LoopVectorizationCostModel(Loop *Lp, ScalarEvolution *Se, + LoopVectorizationLegality *Leg, + const VectorTargetTransformInfo *Vtti): + TheLoop(Lp), SE(Se), Legal(Leg), VTTI(Vtti) { } + + /// Returns the most profitable vectorization factor for the loop that is + /// smaller or equal to the VF argument. This method checks every power + /// of two up to VF. + unsigned findBestVectorizationFactor(unsigned VF = 8); + +private: + /// 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. + unsigned expectedCost(unsigned VF); + + /// Returns the execution time cost of an instruction for a given vector + /// width. Vector width of one means scalar. + unsigned getInstructionCost(Instruction *I, unsigned VF); + + /// A helper function for converting Scalar types to vector types. + /// If the incoming type is void, we return void. If the VF is 1, we return + /// the scalar type. + static Type* ToVectorTy(Type *Scalar, unsigned VF); + + /// The loop that we evaluate. + Loop *TheLoop; + /// Scev analysis. + ScalarEvolution *SE; + + /// Vectorization legality. + LoopVectorizationLegality *Legal; + /// Vector target information. + const VectorTargetTransformInfo *VTTI; +}; + +struct LoopVectorize : public LoopPass { + static char ID; // Pass identification, replacement for typeid + + LoopVectorize() : LoopPass(ID) { + initializeLoopVectorizePass(*PassRegistry::getPassRegistry()); + } + + ScalarEvolution *SE; + DataLayout *DL; + LoopInfo *LI; + TargetTransformInfo *TTI; + DominatorTree *DT; + + virtual bool runOnLoop(Loop *L, LPPassManager &LPM) { + // We only vectorize innermost loops. + if (!L->empty()) + return false; + + SE = &getAnalysis<ScalarEvolution>(); + DL = getAnalysisIfAvailable<DataLayout>(); + LI = &getAnalysis<LoopInfo>(); + TTI = getAnalysisIfAvailable<TargetTransformInfo>(); + DT = &getAnalysis<DominatorTree>(); + + DEBUG(dbgs() << "LV: Checking a loop in \"" << + L->getHeader()->getParent()->getName() << "\"\n"); + + // Check if it is legal to vectorize the loop. + LoopVectorizationLegality LVL(L, SE, DL); + if (!LVL.canVectorize()) { + DEBUG(dbgs() << "LV: Not vectorizing.\n"); + return false; + } + + // Select the preffered vectorization factor. + unsigned VF = 1; + if (VectorizationFactor == 0) { + const VectorTargetTransformInfo *VTTI = 0; + if (TTI) + VTTI = TTI->getVectorTargetTransformInfo(); + // Use the cost model. + LoopVectorizationCostModel CM(L, SE, &LVL, VTTI); + VF = CM.findBestVectorizationFactor(); + + if (VF == 1) { + DEBUG(dbgs() << "LV: Vectorization is possible but not beneficial.\n"); + return false; + } + + } else { + // Use the user command flag. + VF = VectorizationFactor; + } + + DEBUG(dbgs() << "LV: Found a vectorizable loop ("<< VF << ") in "<< + L->getHeader()->getParent()->getParent()->getModuleIdentifier()<< + "\n"); + + // If we decided that it is *legal* to vectorizer the loop then do it. + SingleBlockLoopVectorizer LB(L, SE, LI, DT, &LPM, VF); + LB.vectorize(&LVL); + + DEBUG(verifyFunction(*L->getHeader()->getParent())); + return true; + } + + virtual void getAnalysisUsage(AnalysisUsage &AU) const { + LoopPass::getAnalysisUsage(AU); + AU.addRequiredID(LoopSimplifyID); + AU.addRequiredID(LCSSAID); + AU.addRequired<LoopInfo>(); + AU.addRequired<ScalarEvolution>(); + AU.addRequired<DominatorTree>(); + AU.addPreserved<LoopInfo>(); + AU.addPreserved<DominatorTree>(); + } + +}; + +Value *SingleBlockLoopVectorizer::getBroadcastInstrs(Value *V) { + // Instructions that access the old induction variable + // actually want to get the new one. + if (V == OldInduction) + V = Induction; + // Create the types. + LLVMContext &C = V->getContext(); + Type *VTy = VectorType::get(V->getType(), VF); + Type *I32 = IntegerType::getInt32Ty(C); + Constant *Zero = ConstantInt::get(I32, 0); + Value *Zeros = ConstantAggregateZero::get(VectorType::get(I32, VF)); + Value *UndefVal = UndefValue::get(VTy); + // Insert the value into a new vector. + Value *SingleElem = Builder.CreateInsertElement(UndefVal, V, Zero); + // Broadcast the scalar into all locations in the vector. + Value *Shuf = Builder.CreateShuffleVector(SingleElem, UndefVal, Zeros, + "broadcast"); + // We are accessing the induction variable. Make sure to promote the + // index for each consecutive SIMD lane. This adds 0,1,2 ... to all lanes. + if (V == Induction) + return getConsecutiveVector(Shuf); + return Shuf; +} + +Value *SingleBlockLoopVectorizer::getConsecutiveVector(Value* Val) { + assert(Val->getType()->isVectorTy() && "Must be a vector"); + assert(Val->getType()->getScalarType()->isIntegerTy() && + "Elem must be an integer"); + // Create the types. + Type *ITy = Val->getType()->getScalarType(); + VectorType *Ty = cast<VectorType>(Val->getType()); + unsigned VLen = Ty->getNumElements(); + SmallVector<Constant*, 8> Indices; + + // Create a vector of consecutive numbers from zero to VF. + for (unsigned i = 0; i < VLen; ++i) + Indices.push_back(ConstantInt::get(ITy, i)); + + // Add the consecutive indices to the vector value. + Constant *Cv = ConstantVector::get(Indices); + assert(Cv->getType() == Val->getType() && "Invalid consecutive vec"); + return Builder.CreateAdd(Val, Cv, "induction"); +} + +bool LoopVectorizationLegality::isConsecutiveGep(Value *Ptr) { + GetElementPtrInst *Gep = dyn_cast_or_null<GetElementPtrInst>(Ptr); + if (!Gep) + return false; + + unsigned NumOperands = Gep->getNumOperands(); + Value *LastIndex = Gep->getOperand(NumOperands - 1); + + // Check that all of the gep indices are uniform except for the last. + for (unsigned i = 0; i < NumOperands - 1; ++i) + if (!SE->isLoopInvariant(SE->getSCEV(Gep->getOperand(i)), TheLoop)) + return false; + + // We can emit wide load/stores only of the last index is the induction + // variable. + const SCEV *Last = SE->getSCEV(LastIndex); + if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(Last)) { + const SCEV *Step = AR->getStepRecurrence(*SE); + + // The memory is consecutive because the last index is consecutive + // and all other indices are loop invariant. + if (Step->isOne()) + return true; + } + + return false; +} + +bool LoopVectorizationLegality::isUniform(Value *V) { + return (SE->isLoopInvariant(SE->getSCEV(V), TheLoop)); +} + +Value *SingleBlockLoopVectorizer::getVectorValue(Value *V) { + assert(!V->getType()->isVectorTy() && "Can't widen a vector"); + // If we saved a vectorized copy of V, use it. + Value *&MapEntry = WidenMap[V]; + if (MapEntry) + return MapEntry; + + // Broadcast V and save the value for future uses. + Value *B = getBroadcastInstrs(V); + MapEntry = B; + return B; +} + +Constant* +SingleBlockLoopVectorizer::getUniformVector(unsigned Val, Type* ScalarTy) { + SmallVector<Constant*, 8> Indices; + // Create a vector of consecutive numbers from zero to VF. + for (unsigned i = 0; i < VF; ++i) + Indices.push_back(ConstantInt::get(ScalarTy, Val, true)); + + // Add the consecutive indices to the vector value. + return ConstantVector::get(Indices); +} + +void SingleBlockLoopVectorizer::scalarizeInstruction(Instruction *Instr) { + assert(!Instr->getType()->isAggregateType() && "Can't handle vectors"); + // Holds vector parameters or scalars, in case of uniform vals. + SmallVector<Value*, 8> Params; + + // Find all of the vectorized parameters. + for (unsigned op = 0, e = Instr->getNumOperands(); op != e; ++op) { + Value *SrcOp = Instr->getOperand(op); + + // If we are accessing the old induction variable, use the new one. + if (SrcOp == OldInduction) { + Params.push_back(getBroadcastInstrs(Induction)); + continue; + } + + // Try using previously calculated values. + Instruction *SrcInst = dyn_cast<Instruction>(SrcOp); + + // If the src is an instruction that appeared earlier in the basic block + // then it should already be vectorized. + if (SrcInst && SrcInst->getParent() == Instr->getParent()) { + assert(WidenMap.count(SrcInst) && "Source operand is unavailable"); + // The parameter is a vector value from earlier. + Params.push_back(WidenMap[SrcInst]); + } else { + // The parameter is a scalar from outside the loop. Maybe even a constant. + Params.push_back(SrcOp); + } + } + + assert(Params.size() == Instr->getNumOperands() && + "Invalid number of operands"); + + // Does this instruction return a value ? + bool IsVoidRetTy = Instr->getType()->isVoidTy(); + Value *VecResults = 0; + + // If we have a return value, create an empty vector. We place the scalarized + // instructions in this vector. + if (!IsVoidRetTy) + VecResults = UndefValue::get(VectorType::get(Instr->getType(), VF)); + + // For each scalar that we create: + for (unsigned i = 0; i < VF; ++i) { + Instruction *Cloned = Instr->clone(); + if (!IsVoidRetTy) + Cloned->setName(Instr->getName() + ".cloned"); + // Replace the operands of the cloned instrucions with extracted scalars. + for (unsigned op = 0, e = Instr->getNumOperands(); op != e; ++op) { + Value *Op = Params[op]; + // Param is a vector. Need to extract the right lane. + if (Op->getType()->isVectorTy()) + Op = Builder.CreateExtractElement(Op, Builder.getInt32(i)); + Cloned->setOperand(op, Op); + } + + // Place the cloned scalar in the new loop. + Builder.Insert(Cloned); + + // If the original scalar returns a value we need to place it in a vector + // so that future users will be able to use it. + if (!IsVoidRetTy) + VecResults = Builder.CreateInsertElement(VecResults, Cloned, + Builder.getInt32(i)); + } + + if (!IsVoidRetTy) + WidenMap[Instr] = VecResults; +} + +void +SingleBlockLoopVectorizer::createEmptyLoop(LoopVectorizationLegality *Legal) { + /* + In this function we generate a new loop. The new loop will contain + the vectorized instructions while the old loop will continue to run the + scalar remainder. + + [ ] <-- vector loop bypass. + / | + / v +| [ ] <-- vector pre header. +| | +| v +| [ ] \ +| [ ]_| <-- vector loop. +| | + \ v + >[ ] <--- middle-block. + / | + / v +| [ ] <--- new preheader. +| | +| v +| [ ] \ +| [ ]_| <-- old scalar loop to handle remainder. + \ | + \ v + >[ ] <-- exit block. + ... + */ + + OldInduction = Legal->getInduction(); + assert(OldInduction && "We must have a single phi node."); + Type *IdxTy = OldInduction->getType(); + + // Find the loop boundaries. + const SCEV *ExitCount = SE->getExitCount(OrigLoop, OrigLoop->getHeader()); + assert(ExitCount != SE->getCouldNotCompute() && "Invalid loop count"); + + // Get the total trip count from the count by adding 1. + ExitCount = SE->getAddExpr(ExitCount, + SE->getConstant(ExitCount->getType(), 1)); + // We may need to extend the index in case there is a type mismatch. + // We know that the count starts at zero and does not overflow. + // We are using Zext because it should be less expensive. + if (ExitCount->getType() != IdxTy) + ExitCount = SE->getZeroExtendExpr(ExitCount, IdxTy); + + // This is the original scalar-loop preheader. + BasicBlock *BypassBlock = OrigLoop->getLoopPreheader(); + BasicBlock *ExitBlock = OrigLoop->getExitBlock(); + assert(ExitBlock && "Must have an exit block"); + + // The loop index does not have to start at Zero. It starts with this value. + Value *StartIdx = OldInduction->getIncomingValueForBlock(BypassBlock); + + assert(OrigLoop->getNumBlocks() == 1 && "Invalid loop"); + assert(BypassBlock && "Invalid loop structure"); + + BasicBlock *VectorPH = + BypassBlock->splitBasicBlock(BypassBlock->getTerminator(), "vector.ph"); + BasicBlock *VecBody = VectorPH->splitBasicBlock(VectorPH->getTerminator(), + "vector.body"); + + BasicBlock *MiddleBlock = VecBody->splitBasicBlock(VecBody->getTerminator(), + "middle.block"); + BasicBlock *ScalarPH = + MiddleBlock->splitBasicBlock(MiddleBlock->getTerminator(), + "scalar.preheader"); + // Find the induction variable. + BasicBlock *OldBasicBlock = OrigLoop->getHeader(); + + // Use this IR builder to create the loop instructions (Phi, Br, Cmp) + // inside the loop. + Builder.SetInsertPoint(VecBody->getFirstInsertionPt()); + + // Generate the induction variable. + Induction = Builder.CreatePHI(IdxTy, 2, "index"); + Constant *Step = ConstantInt::get(IdxTy, VF); + + // Expand the trip count and place the new instructions in the preheader. + // Notice that the pre-header does not change, only the loop body. + SCEVExpander Exp(*SE, "induction"); + Instruction *Loc = BypassBlock->getTerminator(); + + // Count holds the overall loop count (N). + Value *Count = Exp.expandCodeFor(ExitCount, Induction->getType(), Loc); + + // Add the start index to the loop count to get the new end index. + Value *IdxEnd = BinaryOperator::CreateAdd(Count, StartIdx, "end.idx", Loc); + + // Now we need to generate the expression for N - (N % VF), which is + // the part that the vectorized body will execute. + Constant *CIVF = ConstantInt::get(IdxTy, VF); + Value *R = BinaryOperator::CreateURem(Count, CIVF, "n.mod.vf", Loc); + Value *CountRoundDown = BinaryOperator::CreateSub(Count, R, "n.vec", Loc); + Value *IdxEndRoundDown = BinaryOperator::CreateAdd(CountRoundDown, StartIdx, + "end.idx.rnd.down", Loc); + + // Now, compare the new count to zero. If it is zero, jump to the scalar part. + Value *Cmp = CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, + IdxEndRoundDown, + StartIdx, + "cmp.zero", Loc); + + LoopVectorizationLegality::RuntimePointerCheck *PtrRtCheck = + Legal->getRuntimePointerCheck(); + Value *MemoryRuntimeCheck = 0; + if (PtrRtCheck->Need) { + unsigned NumPointers = PtrRtCheck->Pointers.size(); + SmallVector<Value* , 2> Starts; + SmallVector<Value* , 2> Ends; + + // Use this type for pointer arithmetic. + Type* PtrArithTy = PtrRtCheck->Pointers[0]->getType(); + + for (unsigned i=0; i < NumPointers; ++i) { + Value *Ptr = PtrRtCheck->Pointers[i]; + const SCEV *Sc = SE->getSCEV(Ptr); + + if (SE->isLoopInvariant(Sc, OrigLoop)) { + DEBUG(dbgs() << "LV1: Adding RT check for a loop invariant ptr:" << + *Ptr <<"\n"); + Starts.push_back(Ptr); + Ends.push_back(Ptr); + } else { + DEBUG(dbgs() << "LV: Adding RT check for range:" << *Ptr <<"\n"); + const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(Sc); + Value *Start = Exp.expandCodeFor(AR->getStart(), PtrArithTy, Loc); + const SCEV *Ex = SE->getExitCount(OrigLoop, OrigLoop->getHeader()); + const SCEV *ScEnd = AR->evaluateAtIteration(Ex, *SE); + assert(!isa<SCEVCouldNotCompute>(ScEnd) && "Invalid scev range."); + Value *End = Exp.expandCodeFor(ScEnd, PtrArithTy, Loc); + Starts.push_back(Start); + Ends.push_back(End); + } + } + + for (unsigned i=0; i < NumPointers; ++i) { + for (unsigned j=i+1; j < NumPointers; ++j) { + Value *Cmp0 = CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_ULE, + Starts[0], Ends[1], "bound0", Loc); + Value *Cmp1 = CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_ULE, + Starts[1], Ends[0], "bound1", Loc); + Value *IsConflict = BinaryOperator::Create(Instruction::And, Cmp0, Cmp1, + "found.conflict", Loc); + if (MemoryRuntimeCheck) { + MemoryRuntimeCheck = BinaryOperator::Create(Instruction::Or, + MemoryRuntimeCheck, + IsConflict, + "conflict.rdx", Loc); + } else { + MemoryRuntimeCheck = IsConflict; + } + } + } + }// end of need-runtime-check code. + + // If we are using memory runtime checks, include them in. + if (MemoryRuntimeCheck) { + Cmp = BinaryOperator::Create(Instruction::Or, Cmp, MemoryRuntimeCheck, + "CntOrMem", Loc); + } + + BranchInst::Create(MiddleBlock, VectorPH, Cmp, Loc); + // Remove the old terminator. + Loc->eraseFromParent(); + + // We are going to resume the execution of the scalar loop. + // This PHI decides on what number to start. If we come from the + // vector loop then we need to start with the end index minus the + // index modulo VF. If we come from a bypass edge then we need to start + // from the real start. + PHINode* ResumeIndex = PHINode::Create(IdxTy, 2, "resume.idx", + MiddleBlock->getTerminator()); + ResumeIndex->addIncoming(StartIdx, BypassBlock); + ResumeIndex->addIncoming(IdxEndRoundDown, VecBody); + + // Add a check in the middle block to see if we have completed + // all of the iterations in the first vector loop. + // If (N - N%VF) == N, then we *don't* need to run the remainder. + Value *CmpN = CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, IdxEnd, + ResumeIndex, "cmp.n", + MiddleBlock->getTerminator()); + + BranchInst::Create(ExitBlock, ScalarPH, CmpN, MiddleBlock->getTerminator()); + // Remove the old terminator. + MiddleBlock->getTerminator()->eraseFromParent(); + + // Create i+1 and fill the PHINode. + Value *NextIdx = Builder.CreateAdd(Induction, Step, "index.next"); + Induction->addIncoming(StartIdx, VectorPH); + Induction->addIncoming(NextIdx, VecBody); + // Create the compare. + Value *ICmp = Builder.CreateICmpEQ(NextIdx, IdxEndRoundDown); + Builder.CreateCondBr(ICmp, MiddleBlock, VecBody); + + // Now we have two terminators. Remove the old one from the block. + VecBody->getTerminator()->eraseFromParent(); + + // Fix the scalar body iteration count. + unsigned BlockIdx = OldInduction->getBasicBlockIndex(ScalarPH); + OldInduction->setIncomingValue(BlockIdx, ResumeIndex); + + // Get ready to start creating new instructions into the vectorized body. + Builder.SetInsertPoint(VecBody->getFirstInsertionPt()); + + // Register the new loop. + Loop* Lp = new Loop(); + LPM->insertLoop(Lp, OrigLoop->getParentLoop()); + + Lp->addBasicBlockToLoop(VecBody, LI->getBase()); + + Loop *ParentLoop = OrigLoop->getParentLoop(); + if (ParentLoop) { + ParentLoop->addBasicBlockToLoop(ScalarPH, LI->getBase()); + ParentLoop->addBasicBlockToLoop(VectorPH, LI->getBase()); + ParentLoop->addBasicBlockToLoop(MiddleBlock, LI->getBase()); + } + + // Save the state. + LoopVectorPreHeader = VectorPH; + LoopScalarPreHeader = ScalarPH; + LoopMiddleBlock = MiddleBlock; + LoopExitBlock = ExitBlock; + LoopVectorBody = VecBody; + LoopScalarBody = OldBasicBlock; + LoopBypassBlock = BypassBlock; +} + +/// This function returns the identity element (or neutral element) for +/// the operation K. +static unsigned +getReductionIdentity(LoopVectorizationLegality::ReductionKind K) { + switch (K) { + case LoopVectorizationLegality::IntegerXor: + case LoopVectorizationLegality::IntegerAdd: + case LoopVectorizationLegality::IntegerOr: + // Adding, Xoring, Oring zero to a number does not change it. + return 0; + case LoopVectorizationLegality::IntegerMult: + // Multiplying a number by 1 does not change it. + return 1; + case LoopVectorizationLegality::IntegerAnd: + // AND-ing a number with an all-1 value does not change it. + return -1; + default: + llvm_unreachable("Unknown reduction kind"); + } +} + +void +SingleBlockLoopVectorizer::vectorizeLoop(LoopVectorizationLegality *Legal) { + //===------------------------------------------------===// + // + // Notice: any optimization or new instruction that go + // into the code below should be also be implemented in + // the cost-model. + // + //===------------------------------------------------===// + typedef SmallVector<PHINode*, 4> PhiVector; + BasicBlock &BB = *OrigLoop->getHeader(); + Constant *Zero = ConstantInt::get( + IntegerType::getInt32Ty(BB.getContext()), 0); + + // In order to support reduction variables we need to be able to vectorize + // Phi nodes. Phi nodes have cycles, so we need to vectorize them in two + // steages. First, we create a new vector PHI node with no incoming edges. + // We use this value when we vectorize all of the instructions that use the + // PHI. Next, after all of the instructions in the block are complete we + // add the new incoming edges to the PHI. At this point all of the + // instructions in the basic block are vectorized, so we can use them to + // construct the PHI. + PhiVector PHIsToFix; + + // For each instruction in the old loop. + for (BasicBlock::iterator it = BB.begin(), e = BB.end(); it != e; ++it) { + Instruction *Inst = it; + + switch (Inst->getOpcode()) { + case Instruction::Br: + // Nothing to do for PHIs and BR, since we already took care of the + // loop control flow instructions. + continue; + case Instruction::PHI:{ + PHINode* P = cast<PHINode>(Inst); + // Special handling for the induction var. + if (OldInduction == Inst) + continue; + // This is phase one of vectorizing PHIs. + // This has to be a reduction variable. + assert(Legal->getReductionVars()->count(P) && "Not a Reduction"); + Type *VecTy = VectorType::get(Inst->getType(), VF); + WidenMap[Inst] = Builder.CreatePHI(VecTy, 2, "vec.phi"); + PHIsToFix.push_back(P); + continue; + } + case Instruction::Add: + case Instruction::FAdd: + case Instruction::Sub: + case Instruction::FSub: + case Instruction::Mul: + case Instruction::FMul: + case Instruction::UDiv: + case Instruction::SDiv: + case Instruction::FDiv: + case Instruction::URem: + case Instruction::SRem: + case Instruction::FRem: + case Instruction::Shl: + case Instruction::LShr: + case Instruction::AShr: + case Instruction::And: + case Instruction::Or: + case Instruction::Xor: { + // Just widen binops. + BinaryOperator *BinOp = dyn_cast<BinaryOperator>(Inst); + Value *A = getVectorValue(Inst->getOperand(0)); + Value *B = getVectorValue(Inst->getOperand(1)); + + // Use this vector value for all users of the original instruction. + Value *V = Builder.CreateBinOp(BinOp->getOpcode(), A, B); + WidenMap[Inst] = V; + + // Update the NSW, NUW and Exact flags. + BinaryOperator *VecOp = cast<BinaryOperator>(V); + if (isa<OverflowingBinaryOperator>(BinOp)) { + VecOp->setHasNoSignedWrap(BinOp->hasNoSignedWrap()); + VecOp->setHasNoUnsignedWrap(BinOp->hasNoUnsignedWrap()); + } + if (isa<PossiblyExactOperator>(VecOp)) + VecOp->setIsExact(BinOp->isExact()); + break; + } + case Instruction::Select: { + // Widen selects. + // If the selector is loop invariant we can create a select + // instruction with a scalar condition. Otherwise, use vector-select. + Value *Cond = Inst->getOperand(0); + bool InvariantCond = SE->isLoopInvariant(SE->getSCEV(Cond), OrigLoop); + + // The condition can be loop invariant but still defined inside the + // loop. This means that we can't just use the original 'cond' value. + // We have to take the 'vectorized' value and pick the first lane. + // Instcombine will make this a no-op. + Cond = getVectorValue(Cond); + if (InvariantCond) + Cond = Builder.CreateExtractElement(Cond, Builder.getInt32(0)); + + Value *Op0 = getVectorValue(Inst->getOperand(1)); + Value *Op1 = getVectorValue(Inst->getOperand(2)); + WidenMap[Inst] = Builder.CreateSelect(Cond, Op0, Op1); + break; + } + + case Instruction::ICmp: + case Instruction::FCmp: { + // Widen compares. Generate vector compares. + bool FCmp = (Inst->getOpcode() == Instruction::FCmp); + CmpInst *Cmp = dyn_cast<CmpInst>(Inst); + Value *A = getVectorValue(Inst->getOperand(0)); + Value *B = getVectorValue(Inst->getOperand(1)); + if (FCmp) + WidenMap[Inst] = Builder.CreateFCmp(Cmp->getPredicate(), A, B); + else + WidenMap[Inst] = Builder.CreateICmp(Cmp->getPredicate(), A, B); + break; + } + + case Instruction::Store: { + // Attempt to issue a wide store. + StoreInst *SI = dyn_cast<StoreInst>(Inst); + Type *StTy = VectorType::get(SI->getValueOperand()->getType(), VF); + Value *Ptr = SI->getPointerOperand(); + unsigned Alignment = SI->getAlignment(); + + assert(!Legal->isUniform(Ptr) && + "We do not allow storing to uniform addresses"); + + GetElementPtrInst *Gep = dyn_cast<GetElementPtrInst>(Ptr); + + // This store does not use GEPs. + if (!Legal->isConsecutiveGep(Gep)) { + scalarizeInstruction(Inst); + break; + } + + // The last index does not have to be the induction. It can be + // consecutive and be a function of the index. For example A[I+1]; + unsigned NumOperands = Gep->getNumOperands(); + Value *LastIndex = getVectorValue(Gep->getOperand(NumOperands - 1)); + LastIndex = Builder.CreateExtractElement(LastIndex, Zero); + + // Create the new GEP with the new induction variable. + GetElementPtrInst *Gep2 = cast<GetElementPtrInst>(Gep->clone()); + Gep2->setOperand(NumOperands - 1, LastIndex); + Ptr = Builder.Insert(Gep2); + Ptr = Builder.CreateBitCast(Ptr, StTy->getPointerTo()); + Value *Val = getVectorValue(SI->getValueOperand()); + Builder.CreateStore(Val, Ptr)->setAlignment(Alignment); + break; + } + case Instruction::Load: { + // Attempt to issue a wide load. + LoadInst *LI = dyn_cast<LoadInst>(Inst); + Type *RetTy = VectorType::get(LI->getType(), VF); + Value *Ptr = LI->getPointerOperand(); + unsigned Alignment = LI->getAlignment(); + GetElementPtrInst *Gep = dyn_cast<GetElementPtrInst>(Ptr); + + // If we don't have a gep, or that the pointer is loop invariant, + // scalarize the load. + if (!Gep || Legal->isUniform(Gep) || !Legal->isConsecutiveGep(Gep)) { + scalarizeInstruction(Inst); + break; + } + + // The last index does not have to be the induction. It can be + // consecutive and be a function of the index. For example A[I+1]; + unsigned NumOperands = Gep->getNumOperands(); + Value *LastIndex = getVectorValue(Gep->getOperand(NumOperands -1)); + LastIndex = Builder.CreateExtractElement(LastIndex, Zero); + + // Create the new GEP with the new induction variable. + GetElementPtrInst *Gep2 = cast<GetElementPtrInst>(Gep->clone()); + Gep2->setOperand(NumOperands - 1, LastIndex); + Ptr = Builder.Insert(Gep2); + Ptr = Builder.CreateBitCast(Ptr, RetTy->getPointerTo()); + LI = Builder.CreateLoad(Ptr); + LI->setAlignment(Alignment); + // Use this vector value for all users of the load. + WidenMap[Inst] = LI; + break; + } + 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: + case Instruction::BitCast: { + /// Vectorize bitcasts. + CastInst *CI = dyn_cast<CastInst>(Inst); + Value *A = getVectorValue(Inst->getOperand(0)); + Type *DestTy = VectorType::get(CI->getType()->getScalarType(), VF); + WidenMap[Inst] = Builder.CreateCast(CI->getOpcode(), A, DestTy); + break; + } + + default: + /// All other instructions are unsupported. Scalarize them. + scalarizeInstruction(Inst); + break; + }// end of switch. + }// end of for_each instr. + + // At this point every instruction in the original loop is widended to + // a vector form. We are almost done. Now, we need to fix the PHI nodes + // that we vectorized. The PHI nodes are currently empty because we did + // not want to introduce cycles. Notice that the remaining PHI nodes + // that we need to fix are reduction variables. + + // Create the 'reduced' values for each of the induction vars. + // The reduced values are the vector values that we scalarize and combine + // after the loop is finished. + for (PhiVector::iterator it = PHIsToFix.begin(), e = PHIsToFix.end(); + it != e; ++it) { + PHINode *RdxPhi = *it; + PHINode *VecRdxPhi = dyn_cast<PHINode>(WidenMap[RdxPhi]); + assert(RdxPhi && "Unable to recover vectorized PHI"); + + // Find the reduction variable descriptor. + assert(Legal->getReductionVars()->count(RdxPhi) && + "Unable to find the reduction variable"); + LoopVectorizationLegality::ReductionDescriptor RdxDesc = + (*Legal->getReductionVars())[RdxPhi]; + + // We need to generate a reduction vector from the incoming scalar. + // To do so, we need to generate the 'identity' vector and overide + // one of the elements with the incoming scalar reduction. We need + // to do it in the vector-loop preheader. + Builder.SetInsertPoint(LoopBypassBlock->getTerminator()); + + // This is the vector-clone of the value that leaves the loop. + Value *VectorExit = getVectorValue(RdxDesc.LoopExitInstr); + Type *VecTy = VectorExit->getType(); + + // Find the reduction identity variable. Zero for addition, or, xor, + // one for multiplication, -1 for And. + Constant *Identity = getUniformVector(getReductionIdentity(RdxDesc.Kind), + VecTy->getScalarType()); + + // This vector is the Identity vector where the first element is the + // incoming scalar reduction. + Value *VectorStart = Builder.CreateInsertElement(Identity, + RdxDesc.StartValue, Zero); + + + // Fix the vector-loop phi. + // We created the induction variable so we know that the + // preheader is the first entry. + BasicBlock *VecPreheader = Induction->getIncomingBlock(0); + + // Reductions do not have to start at zero. They can start with + // any loop invariant values. + VecRdxPhi->addIncoming(VectorStart, VecPreheader); + unsigned SelfEdgeIdx = (RdxPhi)->getBasicBlockIndex(LoopScalarBody); + Value *Val = getVectorValue(RdxPhi->getIncomingValue(SelfEdgeIdx)); + VecRdxPhi->addIncoming(Val, LoopVectorBody); + + // Before each round, move the insertion point right between + // the PHIs and the values we are going to write. + // This allows us to write both PHINodes and the extractelement + // instructions. + Builder.SetInsertPoint(LoopMiddleBlock->getFirstInsertionPt()); + + // This PHINode contains the vectorized reduction variable, or + // the initial value vector, if we bypass the vector loop. + PHINode *NewPhi = Builder.CreatePHI(VecTy, 2, "rdx.vec.exit.phi"); + NewPhi->addIncoming(VectorStart, LoopBypassBlock); + NewPhi->addIncoming(getVectorValue(RdxDesc.LoopExitInstr), LoopVectorBody); + + // Extract the first scalar. + Value *Scalar0 = + Builder.CreateExtractElement(NewPhi, Builder.getInt32(0)); + // Extract and reduce the remaining vector elements. + for (unsigned i=1; i < VF; ++i) { + Value *Scalar1 = + Builder.CreateExtractElement(NewPhi, Builder.getInt32(i)); + switch (RdxDesc.Kind) { + case LoopVectorizationLegality::IntegerAdd: + Scalar0 = Builder.CreateAdd(Scalar0, Scalar1); + break; + case LoopVectorizationLegality::IntegerMult: + Scalar0 = Builder.CreateMul(Scalar0, Scalar1); + break; + case LoopVectorizationLegality::IntegerOr: + Scalar0 = Builder.CreateOr(Scalar0, Scalar1); + break; + case LoopVectorizationLegality::IntegerAnd: + Scalar0 = Builder.CreateAnd(Scalar0, Scalar1); + break; + case LoopVectorizationLegality::IntegerXor: + Scalar0 = Builder.CreateXor(Scalar0, Scalar1); + break; + default: + llvm_unreachable("Unknown reduction operation"); + } + } + + // Now, we need to fix the users of the reduction variable + // inside and outside of the scalar remainder loop. + // We know that the loop is in LCSSA form. We need to update the + // PHI nodes in the exit blocks. + for (BasicBlock::iterator LEI = LoopExitBlock->begin(), + LEE = LoopExitBlock->end(); LEI != LEE; ++LEI) { + PHINode *LCSSAPhi = dyn_cast<PHINode>(LEI); + if (!LCSSAPhi) continue; + + // All PHINodes need to have a single entry edge, or two if + // we already fixed them. + assert(LCSSAPhi->getNumIncomingValues() < 3 && "Invalid LCSSA PHI"); + + // We found our reduction value exit-PHI. Update it with the + // incoming bypass edge. + if (LCSSAPhi->getIncomingValue(0) == RdxDesc.LoopExitInstr) { + // Add an edge coming from the bypass. + LCSSAPhi->addIncoming(Scalar0, LoopMiddleBlock); + break; + } + }// end of the LCSSA phi scan. + + // Fix the scalar loop reduction variable with the incoming reduction sum + // from the vector body and from the backedge value. + int IncomingEdgeBlockIdx = (RdxPhi)->getBasicBlockIndex(LoopScalarBody); + int SelfEdgeBlockIdx = (IncomingEdgeBlockIdx ? 0 : 1); // The other block. + (RdxPhi)->setIncomingValue(SelfEdgeBlockIdx, Scalar0); + (RdxPhi)->setIncomingValue(IncomingEdgeBlockIdx, RdxDesc.LoopExitInstr); + }// end of for each redux variable. +} + +void SingleBlockLoopVectorizer::updateAnalysis() { + // The original basic block. + SE->forgetLoop(OrigLoop); + + // Update the dominator tree information. + assert(DT->properlyDominates(LoopBypassBlock, LoopExitBlock) && + "Entry does not dominate exit."); + + DT->addNewBlock(LoopVectorPreHeader, LoopBypassBlock); + DT->addNewBlock(LoopVectorBody, LoopVectorPreHeader); + DT->addNewBlock(LoopMiddleBlock, LoopBypassBlock); + DT->addNewBlock(LoopScalarPreHeader, LoopMiddleBlock); + DT->changeImmediateDominator(LoopScalarBody, LoopScalarPreHeader); + DT->changeImmediateDominator(LoopExitBlock, LoopMiddleBlock); + + DEBUG(DT->verifyAnalysis()); +} + +bool LoopVectorizationLegality::canVectorize() { + if (!TheLoop->getLoopPreheader()) { + assert(false && "No preheader!!"); + DEBUG(dbgs() << "LV: Loop not normalized." << "\n"); + return false; + } + + // We can only vectorize single basic block loops. + unsigned NumBlocks = TheLoop->getNumBlocks(); + if (NumBlocks != 1) { + DEBUG(dbgs() << "LV: Too many blocks:" << NumBlocks << "\n"); + return false; + } + + // We need to have a loop header. + BasicBlock *BB = TheLoop->getHeader(); + DEBUG(dbgs() << "LV: Found a loop: " << BB->getName() << "\n"); + + // ScalarEvolution needs to be able to find the exit count. + const SCEV *ExitCount = SE->getExitCount(TheLoop, BB); + if (ExitCount == SE->getCouldNotCompute()) { + DEBUG(dbgs() << "LV: SCEV could not compute the loop exit count.\n"); + return false; + } + + // Do not loop-vectorize loops with a tiny trip count. + unsigned TC = SE->getSmallConstantTripCount(TheLoop, BB); + if (TC > 0u && TC < TinyTripCountThreshold) { + DEBUG(dbgs() << "LV: Found a loop with a very small trip count. " << + "This loop is not worth vectorizing.\n"); + return false; + } + + // Go over each instruction and look at memory deps. + if (!canVectorizeBlock(*BB)) { + DEBUG(dbgs() << "LV: Can't vectorize this loop header\n"); + return false; + } + + DEBUG(dbgs() << "LV: We can vectorize this loop" << + (PtrRtCheck.Need ? " (with a runtime bound check)" : "") + <<"!\n"); + + // Okay! We can vectorize. At this point we don't have any other mem analysis + // which may limit our maximum vectorization factor, so just return true with + // no restrictions. + return true; +} + +bool LoopVectorizationLegality::canVectorizeBlock(BasicBlock &BB) { + // Scan the instructions in the block and look for hazards. + for (BasicBlock::iterator it = BB.begin(), e = BB.end(); it != e; ++it) { + Instruction *I = it; + + PHINode *Phi = dyn_cast<PHINode>(I); + if (Phi) { + // This should not happen because the loop should be normalized. + if (Phi->getNumIncomingValues() != 2) { + DEBUG(dbgs() << "LV: Found an invalid PHI.\n"); + return false; + } + // We only look at integer phi nodes. + if (!Phi->getType()->isIntegerTy()) { + DEBUG(dbgs() << "LV: Found an non-int PHI.\n"); + return false; + } + + if (isInductionVariable(Phi)) { + if (Induction) { + DEBUG(dbgs() << "LV: Found too many inductions."<< *Phi <<"\n"); + return false; + } + DEBUG(dbgs() << "LV: Found the induction PHI."<< *Phi <<"\n"); + Induction = Phi; + continue; + } + if (AddReductionVar(Phi, IntegerAdd)) { + DEBUG(dbgs() << "LV: Found an ADD reduction PHI."<< *Phi <<"\n"); + continue; + } + if (AddReductionVar(Phi, IntegerMult)) { + DEBUG(dbgs() << "LV: Found a MUL reduction PHI."<< *Phi <<"\n"); + continue; + } + if (AddReductionVar(Phi, IntegerOr)) { + DEBUG(dbgs() << "LV: Found an OR reduction PHI."<< *Phi <<"\n"); + continue; + } + if (AddReductionVar(Phi, IntegerAnd)) { + DEBUG(dbgs() << "LV: Found an AND reduction PHI."<< *Phi <<"\n"); + continue; + } + if (AddReductionVar(Phi, IntegerXor)) { + DEBUG(dbgs() << "LV: Found a XOR reduction PHI."<< *Phi <<"\n"); + continue; + } + + DEBUG(dbgs() << "LV: Found an unidentified PHI."<< *Phi <<"\n"); + return false; + }// end of PHI handling + + // We still don't handle functions. + CallInst *CI = dyn_cast<CallInst>(I); + if (CI) { + DEBUG(dbgs() << "LV: Found a call site.\n"); + return false; + } + + // We do not re-vectorize vectors. + if (!VectorType::isValidElementType(I->getType()) && + !I->getType()->isVoidTy()) { + DEBUG(dbgs() << "LV: Found unvectorizable type." << "\n"); + return false; + } + + // Reduction instructions are allowed to have exit users. + // All other instructions must not have external users. + if (!AllowedExit.count(I)) + //Check that all of the users of the loop are inside the BB. + for (Value::use_iterator it = I->use_begin(), e = I->use_end(); + it != e; ++it) { + Instruction *U = cast<Instruction>(*it); + // This user may be a reduction exit value. + BasicBlock *Parent = U->getParent(); + if (Parent != &BB) { + DEBUG(dbgs() << "LV: Found an outside user for : "<< *U << "\n"); + return false; + } + } + } // next instr. + + if (!Induction) { + DEBUG(dbgs() << "LV: Did not find an induction var.\n"); + return false; + } + + // Don't vectorize if the memory dependencies do not allow vectorization. + if (!canVectorizeMemory(BB)) + return false; + + // We now know that the loop is vectorizable! + // Collect variables that will remain uniform after vectorization. + std::vector<Value*> Worklist; + + // Start with the conditional branch and walk up the block. + Worklist.push_back(BB.getTerminator()->getOperand(0)); + + while (Worklist.size()) { + Instruction *I = dyn_cast<Instruction>(Worklist.back()); + Worklist.pop_back(); + // Look at instructions inside this block. + if (!I) continue; + if (I->getParent() != &BB) continue; + + // Stop when reaching PHI nodes. + if (isa<PHINode>(I)) { + assert(I == Induction && "Found a uniform PHI that is not the induction"); + break; + } + + // This is a known uniform. + Uniforms.insert(I); + + // Insert all operands. + for (int i=0, Op = I->getNumOperands(); i < Op; ++i) { + Worklist.push_back(I->getOperand(i)); + } + } + + return true; +} + +bool LoopVectorizationLegality::canVectorizeMemory(BasicBlock &BB) { + typedef SmallVector<Value*, 16> ValueVector; + typedef SmallPtrSet<Value*, 16> ValueSet; + // Holds the Load and Store *instructions*. + ValueVector Loads; + ValueVector Stores; + PtrRtCheck.Pointers.clear(); + PtrRtCheck.Need = false; + + // Scan the BB and collect legal loads and stores. + for (BasicBlock::iterator it = BB.begin(), e = BB.end(); it != e; ++it) { + Instruction *I = it; + + // If this is a load, save it. If this instruction can read from memory + // but is not a load, then we quit. Notice that we don't handle function + // calls that read or write. + if (I->mayReadFromMemory()) { + LoadInst *Ld = dyn_cast<LoadInst>(I); + if (!Ld) return false; + if (!Ld->isSimple()) { + DEBUG(dbgs() << "LV: Found a non-simple load.\n"); + return false; + } + Loads.push_back(Ld); + continue; + } + + // Save store instructions. Abort if other instructions write to memory. + if (I->mayWriteToMemory()) { + StoreInst *St = dyn_cast<StoreInst>(I); + if (!St) return false; + if (!St->isSimple()) { + DEBUG(dbgs() << "LV: Found a non-simple store.\n"); + return false; + } + Stores.push_back(St); + } + } // next instr. + + // Now we have two lists that hold the loads and the stores. + // Next, we find the pointers that they use. + + // Check if we see any stores. If there are no stores, then we don't + // care if the pointers are *restrict*. + if (!Stores.size()) { + DEBUG(dbgs() << "LV: Found a read-only loop!\n"); + return true; + } + + // Holds the read and read-write *pointers* that we find. + ValueVector Reads; + ValueVector ReadWrites; + + // Holds the analyzed pointers. We don't want to call GetUnderlyingObjects + // multiple times on the same object. If the ptr is accessed twice, once + // for read and once for write, it will only appear once (on the write + // list). This is okay, since we are going to check for conflicts between + // writes and between reads and writes, but not between reads and reads. + ValueSet Seen; + + ValueVector::iterator I, IE; + for (I = Stores.begin(), IE = Stores.end(); I != IE; ++I) { + StoreInst *ST = dyn_cast<StoreInst>(*I); + assert(ST && "Bad StoreInst"); + Value* Ptr = ST->getPointerOperand(); + + if (isUniform(Ptr)) { + DEBUG(dbgs() << "LV: We don't allow storing to uniform addresses\n"); + return false; + } + + // If we did *not* see this pointer before, insert it to + // the read-write list. At this phase it is only a 'write' list. + if (Seen.insert(Ptr)) + ReadWrites.push_back(Ptr); + } + + for (I = Loads.begin(), IE = Loads.end(); I != IE; ++I) { + LoadInst *LD = dyn_cast<LoadInst>(*I); + assert(LD && "Bad LoadInst"); + Value* Ptr = LD->getPointerOperand(); + // If we did *not* see this pointer before, insert it to the + // read list. If we *did* see it before, then it is already in + // the read-write list. This allows us to vectorize expressions + // such as A[i] += x; Because the address of A[i] is a read-write + // pointer. This only works if the index of A[i] is consecutive. + // If the address of i is unknown (for example A[B[i]]) then we may + // read a few words, modify, and write a few words, and some of the + // words may be written to the same address. + if (Seen.insert(Ptr) || !isConsecutiveGep(Ptr)) + Reads.push_back(Ptr); + } + + // If we write (or read-write) to a single destination and there are no + // other reads in this loop then is it safe to vectorize. + if (ReadWrites.size() == 1 && Reads.size() == 0) { + DEBUG(dbgs() << "LV: Found a write-only loop!\n"); + return true; + } + + // Find pointers with computable bounds. We are going to use this information + // to place a runtime bound check. + bool RT = true; + for (I = ReadWrites.begin(), IE = ReadWrites.end(); I != IE; ++I) + if (hasComputableBounds(*I)) { + PtrRtCheck.Pointers.push_back(*I); + DEBUG(dbgs() << "LV: Found a runtime check ptr:" << **I <<"\n"); + } else { + RT = false; + break; + } + for (I = Reads.begin(), IE = Reads.end(); I != IE; ++I) + if (hasComputableBounds(*I)) { + PtrRtCheck.Pointers.push_back(*I); + DEBUG(dbgs() << "LV: Found a runtime check ptr:" << **I <<"\n"); + } else { + RT = false; + break; + } + + // Check that we did not collect too many pointers or found a + // unsizeable pointer. + if (!RT || PtrRtCheck.Pointers.size() > RuntimeMemoryCheckThreshold) { + PtrRtCheck.Pointers.clear(); + RT = false; + } + + PtrRtCheck.Need = RT; + + if (RT) { + DEBUG(dbgs() << "LV: We can perform a memory runtime check if needed.\n"); + } + + // Now that the pointers are in two lists (Reads and ReadWrites), we + // can check that there are no conflicts between each of the writes and + // between the writes to the reads. + ValueSet WriteObjects; + ValueVector TempObjects; + + // Check that the read-writes do not conflict with other read-write + // pointers. + for (I = ReadWrites.begin(), IE = ReadWrites.end(); I != IE; ++I) { + GetUnderlyingObjects(*I, TempObjects, DL); + for (ValueVector::iterator it=TempObjects.begin(), e=TempObjects.end(); + it != e; ++it) { + if (!isIdentifiedObject(*it)) { + DEBUG(dbgs() << "LV: Found an unidentified write ptr:"<< **it <<"\n"); + return RT; + } + if (!WriteObjects.insert(*it)) { + DEBUG(dbgs() << "LV: Found a possible write-write reorder:" + << **it <<"\n"); + return RT; + } + } + TempObjects.clear(); + } + + /// Check that the reads don't conflict with the read-writes. + for (I = Reads.begin(), IE = Reads.end(); I != IE; ++I) { + GetUnderlyingObjects(*I, TempObjects, DL); + for (ValueVector::iterator it=TempObjects.begin(), e=TempObjects.end(); + it != e; ++it) { + if (!isIdentifiedObject(*it)) { + DEBUG(dbgs() << "LV: Found an unidentified read ptr:"<< **it <<"\n"); + return RT; + } + if (WriteObjects.count(*it)) { + DEBUG(dbgs() << "LV: Found a possible read/write reorder:" + << **it <<"\n"); + return RT; + } + } + TempObjects.clear(); + } + + // It is safe to vectorize and we don't need any runtime checks. + DEBUG(dbgs() << "LV: We don't need a runtime memory check.\n"); + PtrRtCheck.Pointers.clear(); + PtrRtCheck.Need = false; + return true; +} + +bool LoopVectorizationLegality::AddReductionVar(PHINode *Phi, + ReductionKind Kind) { + if (Phi->getNumIncomingValues() != 2) + return false; + + // Find the possible incoming reduction variable. + BasicBlock *BB = Phi->getParent(); + int SelfEdgeIdx = Phi->getBasicBlockIndex(BB); + int InEdgeBlockIdx = (SelfEdgeIdx ? 0 : 1); // The other entry. + Value *RdxStart = Phi->getIncomingValue(InEdgeBlockIdx); + + // ExitInstruction is the single value which is used outside the loop. + // We only allow for a single reduction value to be used outside the loop. + // This includes users of the reduction, variables (which form a cycle + // which ends in the phi node). + Instruction *ExitInstruction = 0; + + // Iter is our iterator. We start with the PHI node and scan for all of the + // users of this instruction. All users must be instructions which can be + // used as reduction variables (such as ADD). We may have a single + // out-of-block user. They cycle must end with the original PHI. + // Also, we can't have multiple block-local users. + Instruction *Iter = Phi; + while (true) { + // Any reduction instr must be of one of the allowed kinds. + if (!isReductionInstr(Iter, Kind)) + return false; + + // Did we found a user inside this block ? + bool FoundInBlockUser = false; + // Did we reach the initial PHI node ? + bool FoundStartPHI = false; + + // If the instruction has no users then this is a broken + // chain and can't be a reduction variable. + if (Iter->use_empty()) + return false; + + // For each of the *users* of iter. + for (Value::use_iterator it = Iter->use_begin(), e = Iter->use_end(); + it != e; ++it) { + Instruction *U = cast<Instruction>(*it); + // We already know that the PHI is a user. + if (U == Phi) { + FoundStartPHI = true; + continue; + } + // Check if we found the exit user. + BasicBlock *Parent = U->getParent(); + if (Parent != BB) { + // We must have a single exit instruction. + if (ExitInstruction != 0) + return false; + ExitInstruction = Iter; + } + // We can't have multiple inside users. + if (FoundInBlockUser) + return false; + FoundInBlockUser = true; + Iter = U; + } + + // We found a reduction var if we have reached the original + // phi node and we only have a single instruction with out-of-loop + // users. + if (FoundStartPHI && ExitInstruction) { + // This instruction is allowed to have out-of-loop users. + AllowedExit.insert(ExitInstruction); + + // Save the description of this reduction variable. + ReductionDescriptor RD(RdxStart, ExitInstruction, Kind); + Reductions[Phi] = RD; + return true; + } + } +} + +bool +LoopVectorizationLegality::isReductionInstr(Instruction *I, + ReductionKind Kind) { + switch (I->getOpcode()) { + default: + return false; + case Instruction::PHI: + // possibly. + return true; + case Instruction::Add: + case Instruction::Sub: + return Kind == IntegerAdd; + case Instruction::Mul: + case Instruction::UDiv: + case Instruction::SDiv: + return Kind == IntegerMult; + case Instruction::And: + return Kind == IntegerAnd; + case Instruction::Or: + return Kind == IntegerOr; + case Instruction::Xor: + return Kind == IntegerXor; + } +} + +bool LoopVectorizationLegality::isInductionVariable(PHINode *Phi) { + // Check that the PHI is consecutive and starts at zero. + const SCEV *PhiScev = SE->getSCEV(Phi); + const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(PhiScev); + if (!AR) { + DEBUG(dbgs() << "LV: PHI is not a poly recurrence.\n"); + return false; + } + const SCEV *Step = AR->getStepRecurrence(*SE); + + if (!Step->isOne()) { + DEBUG(dbgs() << "LV: PHI stride does not equal one.\n"); + return false; + } + return true; +} + +bool LoopVectorizationLegality::hasComputableBounds(Value *Ptr) { + const SCEV *PhiScev = SE->getSCEV(Ptr); + const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(PhiScev); + if (!AR) + return false; + + return AR->isAffine(); +} + +unsigned +LoopVectorizationCostModel::findBestVectorizationFactor(unsigned VF) { + if (!VTTI) { + DEBUG(dbgs() << "LV: No vector target information. Not vectorizing. \n"); + return 1; + } + + float Cost = expectedCost(1); + unsigned Width = 1; + DEBUG(dbgs() << "LV: Scalar loop costs: "<< (int)Cost << ".\n"); + for (unsigned i=2; i <= VF; i*=2) { + // Notice that the vector loop needs to be executed less times, so + // we need to divide the cost of the vector loops by the width of + // the vector elements. + float VectorCost = expectedCost(i) / (float)i; + DEBUG(dbgs() << "LV: Vector loop of width "<< i << " costs: " << + (int)VectorCost << ".\n"); + if (VectorCost < Cost) { + Cost = VectorCost; + Width = i; + } + } + + DEBUG(dbgs() << "LV: Selecting VF = : "<< Width << ".\n"); + return Width; +} + +unsigned LoopVectorizationCostModel::expectedCost(unsigned VF) { + // We can only estimate the cost of single basic block loops. + assert(1 == TheLoop->getNumBlocks() && "Too many blocks in loop"); + + BasicBlock *BB = TheLoop->getHeader(); + unsigned Cost = 0; + + // For each instruction in the old loop. + for (BasicBlock::iterator it = BB->begin(), e = BB->end(); it != e; ++it) { + Instruction *Inst = it; + unsigned C = getInstructionCost(Inst, VF); + Cost += C; + DEBUG(dbgs() << "LV: Found an estimated cost of "<< C <<" for VF "<< VF << + " For instruction: "<< *Inst << "\n"); + } + + return Cost; +} + +unsigned +LoopVectorizationCostModel::getInstructionCost(Instruction *I, unsigned VF) { + assert(VTTI && "Invalid vector target transformation info"); + + // If we know that this instruction will remain uniform, check the cost of + // the scalar version. + if (Legal->isUniformAfterVectorization(I)) + VF = 1; + + Type *RetTy = I->getType(); + Type *VectorTy = ToVectorTy(RetTy, VF); + + + // TODO: We need to estimate the cost of intrinsic calls. + switch (I->getOpcode()) { + case Instruction::GetElementPtr: + // We mark this instruction as zero-cost because scalar GEPs are usually + // lowered to the intruction addressing mode. At the moment we don't + // generate vector geps. + return 0; + case Instruction::Br: { + return VTTI->getCFInstrCost(I->getOpcode()); + } + case Instruction::PHI: + return 0; + case Instruction::Add: + case Instruction::FAdd: + case Instruction::Sub: + case Instruction::FSub: + case Instruction::Mul: + case Instruction::FMul: + case Instruction::UDiv: + case Instruction::SDiv: + case Instruction::FDiv: + case Instruction::URem: + case Instruction::SRem: + case Instruction::FRem: + case Instruction::Shl: + case Instruction::LShr: + case Instruction::AShr: + case Instruction::And: + case Instruction::Or: + case Instruction::Xor: { + return VTTI->getArithmeticInstrCost(I->getOpcode(), VectorTy); + } + case Instruction::Select: { + SelectInst *SI = cast<SelectInst>(I); + const SCEV *CondSCEV = SE->getSCEV(SI->getCondition()); + bool ScalarCond = (SE->isLoopInvariant(CondSCEV, TheLoop)); + Type *CondTy = SI->getCondition()->getType(); + if (ScalarCond) + CondTy = VectorType::get(CondTy, VF); + + return VTTI->getCmpSelInstrCost(I->getOpcode(), VectorTy, CondTy); + } + case Instruction::ICmp: + case Instruction::FCmp: { + Type *ValTy = I->getOperand(0)->getType(); + VectorTy = ToVectorTy(ValTy, VF); + return VTTI->getCmpSelInstrCost(I->getOpcode(), VectorTy); + } + case Instruction::Store: { + StoreInst *SI = cast<StoreInst>(I); + Type *ValTy = SI->getValueOperand()->getType(); + VectorTy = ToVectorTy(ValTy, VF); + + if (VF == 1) + return VTTI->getMemoryOpCost(I->getOpcode(), ValTy, + SI->getAlignment(), SI->getPointerAddressSpace()); + + // Scalarized stores. + if (!Legal->isConsecutiveGep(SI->getPointerOperand())) { + unsigned Cost = 0; + unsigned ExtCost = VTTI->getInstrCost(Instruction::ExtractElement, + ValTy); + // The cost of extracting from the value vector. + Cost += VF * (ExtCost); + // The cost of the scalar stores. + Cost += VF * VTTI->getMemoryOpCost(I->getOpcode(), + ValTy->getScalarType(), + SI->getAlignment(), + SI->getPointerAddressSpace()); + return Cost; + } + + // Wide stores. + return VTTI->getMemoryOpCost(I->getOpcode(), VectorTy, SI->getAlignment(), + SI->getPointerAddressSpace()); + } + case Instruction::Load: { + LoadInst *LI = cast<LoadInst>(I); + + if (VF == 1) + return VTTI->getMemoryOpCost(I->getOpcode(), RetTy, + LI->getAlignment(), + LI->getPointerAddressSpace()); + + // Scalarized loads. + if (!Legal->isConsecutiveGep(LI->getPointerOperand())) { + unsigned Cost = 0; + unsigned InCost = VTTI->getInstrCost(Instruction::InsertElement, RetTy); + // The cost of inserting the loaded value into the result vector. + Cost += VF * (InCost); + // The cost of the scalar stores. + Cost += VF * VTTI->getMemoryOpCost(I->getOpcode(), + RetTy->getScalarType(), + LI->getAlignment(), + LI->getPointerAddressSpace()); + return Cost; + } + + // Wide loads. + return VTTI->getMemoryOpCost(I->getOpcode(), VectorTy, LI->getAlignment(), + LI->getPointerAddressSpace()); + } + 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: + case Instruction::BitCast: { + Type *SrcVecTy = ToVectorTy(I->getOperand(0)->getType(), VF); + return VTTI->getCastInstrCost(I->getOpcode(), VectorTy, SrcVecTy); + } + default: { + // We are scalarizing the instruction. Return the cost of the scalar + // instruction, plus the cost of insert and extract into vector + // elements, times the vector width. + unsigned Cost = 0; + + bool IsVoid = RetTy->isVoidTy(); + + unsigned InsCost = (IsVoid ? 0 : + VTTI->getInstrCost(Instruction::InsertElement, + VectorTy)); + + unsigned ExtCost = VTTI->getInstrCost(Instruction::ExtractElement, + VectorTy); + + // The cost of inserting the results plus extracting each one of the + // operands. + Cost += VF * (InsCost + ExtCost * I->getNumOperands()); + + // The cost of executing VF copies of the scalar instruction. + Cost += VF * VTTI->getInstrCost(I->getOpcode(), RetTy); + return Cost; + } + }// end of switch. +} + +Type* LoopVectorizationCostModel::ToVectorTy(Type *Scalar, unsigned VF) { + if (Scalar->isVoidTy() || VF == 1) + return Scalar; + return VectorType::get(Scalar, VF); +} + +} // namespace + +char LoopVectorize::ID = 0; +static const char lv_name[] = "Loop Vectorization"; +INITIALIZE_PASS_BEGIN(LoopVectorize, LV_NAME, lv_name, false, false) +INITIALIZE_AG_DEPENDENCY(AliasAnalysis) +INITIALIZE_PASS_DEPENDENCY(ScalarEvolution) +INITIALIZE_PASS_DEPENDENCY(LoopSimplify) +INITIALIZE_PASS_END(LoopVectorize, LV_NAME, lv_name, false, false) + +namespace llvm { + Pass *createLoopVectorizePass() { + return new LoopVectorize(); + } +} + |
