#include "llvm/Transforms/Vectorize.h"
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/EquivalenceClasses.h"
+#include "llvm/ADT/Hashing.h"
#include "llvm/ADT/MapVector.h"
#include "llvm/ADT/SetVector.h"
#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/StringExtras.h"
#include "llvm/Analysis/AliasAnalysis.h"
-#include "llvm/Analysis/Dominators.h"
+#include "llvm/Analysis/BlockFrequencyInfo.h"
#include "llvm/Analysis/LoopInfo.h"
#include "llvm/Analysis/LoopIterator.h"
#include "llvm/Analysis/LoopPass.h"
#include "llvm/Analysis/ScalarEvolutionExpressions.h"
#include "llvm/Analysis/TargetTransformInfo.h"
#include "llvm/Analysis/ValueTracking.h"
-#include "llvm/Analysis/Verifier.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DataLayout.h"
+#include "llvm/IR/DebugInfo.h"
#include "llvm/IR/DerivedTypes.h"
+#include "llvm/IR/Dominators.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/IRBuilder.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Module.h"
+#include "llvm/IR/PatternMatch.h"
#include "llvm/IR/Type.h"
#include "llvm/IR/Value.h"
+#include "llvm/IR/ValueHandle.h"
+#include "llvm/IR/Verifier.h"
#include "llvm/Pass.h"
+#include "llvm/Support/BranchProbability.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
-#include "llvm/Support/PatternMatch.h"
+#include "llvm/Support/Format.h"
#include "llvm/Support/raw_ostream.h"
-#include "llvm/Support/ValueHandle.h"
#include "llvm/Target/TargetLibraryInfo.h"
#include "llvm/Transforms/Scalar.h"
#include "llvm/Transforms/Utils/BasicBlockUtils.h"
#include "llvm/Transforms/Utils/Local.h"
+#include "llvm/Transforms/Utils/VectorUtils.h"
#include <algorithm>
#include <map>
"trip count that is smaller than this "
"value."));
+/// This enables versioning on the strides of symbolically striding memory
+/// accesses in code like the following.
+/// for (i = 0; i < N; ++i)
+/// A[i * Stride1] += B[i * Stride2] ...
+///
+/// Will be roughly translated to
+/// if (Stride1 == 1 && Stride2 == 1) {
+/// for (i = 0; i < N; i+=4)
+/// A[i:i+3] += ...
+/// } else
+/// ...
+static cl::opt<bool> EnableMemAccessVersioning(
+ "enable-mem-access-versioning", cl::init(true), cl::Hidden,
+ cl::desc("Enable symblic stride memory access versioning"));
+
/// We don't unroll loops with a known constant trip count below this number.
static const unsigned TinyTripCountUnrollThreshold = 128;
/// Maximum simd width.
static const unsigned MaxVectorWidth = 64;
+static cl::opt<unsigned> 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<unsigned> ForceTargetNumVectorRegs(
+ "force-target-num-vector-regs", cl::init(0), cl::Hidden,
+ cl::desc("A flag that overrides the target's number of vector registers."));
+
/// Maximum vectorization unroll count.
static const unsigned MaxUnrollFactor = 16;
+static cl::opt<unsigned> ForceTargetMaxScalarUnrollFactor(
+ "force-target-max-scalar-unroll", cl::init(0), cl::Hidden,
+ cl::desc("A flag that overrides the target's max unroll factor for scalar "
+ "loops."));
+
+static cl::opt<unsigned> ForceTargetMaxVectorUnrollFactor(
+ "force-target-max-vector-unroll", cl::init(0), cl::Hidden,
+ cl::desc("A flag that overrides the target's max unroll factor for "
+ "vectorized loops."));
+
+static cl::opt<unsigned> 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<unsigned> SmallLoopCost(
+ "small-loop-cost", cl::init(20), cl::Hidden,
+ cl::desc("The cost of a loop that is considered 'small' by the unroller."));
+
+static cl::opt<bool> LoopVectorizeWithBlockFrequency(
+ "loop-vectorize-with-block-frequency", cl::init(false), 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 unroll loops for load/store throughput.
+static cl::opt<bool> EnableLoadStoreRuntimeUnroll(
+ "enable-loadstore-runtime-unroll", cl::init(true), cl::Hidden,
+ cl::desc("Enable runtime unrolling until load/store ports are saturated"));
+
+/// The number of stores in a loop that are allowed to need predication.
+static cl::opt<unsigned> NumberOfStoresToPredicate(
+ "vectorize-num-stores-pred", cl::init(1), cl::Hidden,
+ cl::desc("Max number of stores to be predicated behind an if."));
+
+static cl::opt<bool> EnableIndVarRegisterHeur(
+ "enable-ind-var-reg-heur", cl::init(true), cl::Hidden,
+ cl::desc("Count the induction variable only once when unrolling"));
+
+static cl::opt<bool> EnableCondStoresVectorization(
+ "enable-cond-stores-vec", cl::init(false), cl::Hidden,
+ cl::desc("Enable if predication of stores during vectorization."));
+
namespace {
// Forward declarations.
class InnerLoopVectorizer {
public:
InnerLoopVectorizer(Loop *OrigLoop, ScalarEvolution *SE, LoopInfo *LI,
- DominatorTree *DT, DataLayout *DL,
+ DominatorTree *DT, const DataLayout *DL,
const TargetLibraryInfo *TLI, unsigned VecWidth,
unsigned UnrollFactor)
: OrigLoop(OrigLoop), SE(SE), LI(LI), DT(DT), DL(DL), TLI(TLI),
VF(VecWidth), UF(UnrollFactor), Builder(SE->getContext()), Induction(0),
- OldInduction(0), WidenMap(UnrollFactor) {}
+ OldInduction(0), WidenMap(UnrollFactor), Legal(0) {}
// Perform the actual loop widening (vectorization).
- void vectorize(LoopVectorizationLegality *Legal) {
+ void vectorize(LoopVectorizationLegality *L) {
+ Legal = L;
// Create a new empty loop. Unlink the old loop and connect the new one.
- createEmptyLoop(Legal);
+ createEmptyLoop();
// 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);
+ vectorizeLoop();
// Register the new loop and update the analysis passes.
updateAnalysis();
}
-private:
+ virtual ~InnerLoopVectorizer() {}
+
+protected:
/// A small list of PHINodes.
typedef SmallVector<PHINode*, 4> PhiVector;
/// When we unroll loops we have multiple vector values for each scalar.
typedef DenseMap<std::pair<BasicBlock*, BasicBlock*>,
VectorParts> EdgeMaskCache;
- /// Add code that checks at runtime if the accessed arrays overlap.
- /// Returns the comparator value or NULL if no check is needed.
- Instruction *addRuntimeCheck(LoopVectorizationLegality *Legal,
- Instruction *Loc);
+ /// \brief Add code that checks at runtime if the accessed arrays overlap.
+ ///
+ /// Returns a pair of instructions where the first element is the first
+ /// instruction generated in possibly a sequence of instructions and the
+ /// second value is the final comparator value or NULL if no check is needed.
+ std::pair<Instruction *, Instruction *> addRuntimeCheck(Instruction *Loc);
+
+ /// \brief Add checks for strides that where assumed to be 1.
+ ///
+ /// Returns the last check instruction and the first check instruction in the
+ /// pair as (first, last).
+ std::pair<Instruction *, Instruction *> addStrideCheck(Instruction *Loc);
+
/// Create an empty loop, based on the loop ranges of the old loop.
- void createEmptyLoop(LoopVectorizationLegality *Legal);
+ void createEmptyLoop();
/// Copy and widen the instructions from the old loop.
- void vectorizeLoop(LoopVectorizationLegality *Legal);
+ virtual void vectorizeLoop();
+
+ /// \brief The Loop exit block may have single value PHI nodes where the
+ /// incoming value is 'Undef'. While vectorizing we only handled real values
+ /// that were defined inside the loop. Here we fix the 'undef case'.
+ /// See PR14725.
+ void fixLCSSAPHIs();
/// A helper function that computes the predicate of the block BB, assuming
/// that the header block of the loop is set to True. It returns the *entry*
VectorParts createEdgeMask(BasicBlock *Src, BasicBlock *Dst);
/// A helper function to vectorize a single BB within the innermost loop.
- void vectorizeBlockInLoop(LoopVectorizationLegality *Legal, BasicBlock *BB,
- PhiVector *PV);
+ void vectorizeBlockInLoop(BasicBlock *BB, PhiVector *PV);
+
+ /// Vectorize a single PHINode in a block. This method handles the induction
+ /// variable canonicalization. It supports both VF = 1 for unrolled loops and
+ /// arbitrary length vectors.
+ void widenPHIInstruction(Instruction *PN, VectorParts &Entry,
+ unsigned UF, unsigned VF, PhiVector *PV);
/// 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);
+ /// of scalars. If \p IfPredicateStore is true we need to 'hide' each
+ /// scalarized instruction behind an if block predicated on the control
+ /// dependence of the instruction.
+ virtual void scalarizeInstruction(Instruction *Instr,
+ bool IfPredicateStore=false);
/// Vectorize Load and Store instructions,
- void vectorizeMemoryInstruction(Instruction *Instr,
- LoopVectorizationLegality *Legal);
+ virtual void vectorizeMemoryInstruction(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);
+ virtual Value *getBroadcastInstrs(Value *V);
/// This function adds 0, 1, 2 ... to each vector element, starting at zero.
/// If Negate is set then negative numbers are added e.g. (0, -1, -2, ...).
/// The sequence starts at StartIndex.
- Value *getConsecutiveVector(Value* Val, int StartIdx, bool Negate);
+ virtual Value *getConsecutiveVector(Value* Val, int StartIdx, bool Negate);
/// When we go over instructions in the basic block we rely on previous
/// values within the current basic block or on loop invariant values.
VectorParts &getVectorValue(Value *V);
/// Generate a shuffle sequence that will reverse the vector Vec.
- Value *reverseVector(Value *Vec);
+ virtual Value *reverseVector(Value *Vec);
/// This is a helper class that holds the vectorizer state. It maps scalar
/// instructions to vector instructions. When the code is 'unrolled' then
/// Dominator Tree.
DominatorTree *DT;
/// Data Layout.
- DataLayout *DL;
+ const DataLayout *DL;
/// Target Library Info.
const TargetLibraryInfo *TLI;
/// The vectorization SIMD factor to use. Each vector will have this many
/// vector elements.
unsigned VF;
+
+protected:
/// The vectorization unroll factor to use. Each scalar is vectorized to this
/// many different vector instructions.
unsigned UF;
///The ExitBlock of the scalar loop.
BasicBlock *LoopExitBlock;
///The vector loop body.
- BasicBlock *LoopVectorBody;
+ SmallVector<BasicBlock *, 4> LoopVectorBody;
///The scalar loop body.
BasicBlock *LoopScalarBody;
/// A list of all bypass blocks. The first block is the entry of the loop.
/// Maps scalars to widened vectors.
ValueMap WidenMap;
EdgeMaskCache MaskCache;
+
+ LoopVectorizationLegality *Legal;
+};
+
+class InnerLoopUnroller : public InnerLoopVectorizer {
+public:
+ InnerLoopUnroller(Loop *OrigLoop, ScalarEvolution *SE, LoopInfo *LI,
+ DominatorTree *DT, const DataLayout *DL,
+ const TargetLibraryInfo *TLI, unsigned UnrollFactor) :
+ InnerLoopVectorizer(OrigLoop, SE, LI, DT, DL, TLI, 1, UnrollFactor) { }
+
+private:
+ void scalarizeInstruction(Instruction *Instr,
+ bool IfPredicateStore = false) override;
+ void vectorizeMemoryInstruction(Instruction *Instr) override;
+ Value *getBroadcastInstrs(Value *V) override;
+ Value *getConsecutiveVector(Value* Val, int StartIdx, bool Negate) override;
+ Value *reverseVector(Value *Vec) override;
};
/// \brief Look for a meaningful debug location on the instruction or it's
B.SetCurrentDebugLocation(DebugLoc());
}
+#ifndef NDEBUG
+/// \return string containing a file name and a line # for the given
+/// instruction.
+static format_object3<const char *, const char *, unsigned>
+getDebugLocString(const Instruction *I) {
+ if (!I)
+ return format<const char *, const char *, unsigned>("", "", "", 0U);
+ MDNode *N = I->getMetadata("dbg");
+ if (!N) {
+ const StringRef ModuleName =
+ I->getParent()->getParent()->getParent()->getModuleIdentifier();
+ return format<const char *, const char *, unsigned>("%s", ModuleName.data(),
+ "", 0U);
+ }
+ const DILocation Loc(N);
+ const unsigned LineNo = Loc.getLineNumber();
+ const char *DirName = Loc.getDirectory().data();
+ const char *FileName = Loc.getFilename().data();
+ return format("%s/%s:%u", DirName, FileName, LineNo);
+}
+#endif
+
/// 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
/// induction variable and the different reduction variables.
class LoopVectorizationLegality {
public:
- LoopVectorizationLegality(Loop *L, ScalarEvolution *SE, DataLayout *DL,
+ unsigned NumLoads;
+ unsigned NumStores;
+ unsigned NumPredStores;
+
+ LoopVectorizationLegality(Loop *L, ScalarEvolution *SE, const DataLayout *DL,
DominatorTree *DT, TargetLibraryInfo *TLI)
- : TheLoop(L), SE(SE), DL(DL), DT(DT), TLI(TLI),
- Induction(0), WidestIndTy(0), HasFunNoNaNAttr(false),
+ : NumLoads(0), NumStores(0), NumPredStores(0), TheLoop(L), SE(SE), DL(DL),
+ DT(DT), TLI(TLI), Induction(0), WidestIndTy(0), HasFunNoNaNAttr(false),
MaxSafeDepDistBytes(-1U) {}
/// This enum represents the kinds of reductions that we support.
MRK_FloatMax
};
- /// This POD struct holds information about reduction variables.
+ /// This struct holds information about reduction variables.
struct ReductionDescriptor {
ReductionDescriptor() : StartValue(0), LoopExitInstr(0),
Kind(RK_NoReduction), MinMaxKind(MRK_Invalid) {}
MinMaxReductionKind MinMaxKind;
};
- // This POD struct holds information about the memory runtime legality
- // check that a group of pointers do not overlap.
+ /// This struct holds information about the memory runtime legality
+ /// check that a group of pointers do not overlap.
struct RuntimePointerCheck {
RuntimePointerCheck() : Need(false) {}
Pointers.clear();
Starts.clear();
Ends.clear();
+ IsWritePtr.clear();
+ DependencySetId.clear();
}
/// Insert a pointer and calculate the start and end SCEVs.
void insert(ScalarEvolution *SE, Loop *Lp, Value *Ptr, bool WritePtr,
- unsigned DepSetId);
+ unsigned DepSetId, ValueToValueMap &Strides);
/// This flag indicates if we need to add the runtime check.
bool Need;
SmallVector<unsigned, 2> DependencySetId;
};
- /// A POD for saving information about induction variables.
+ /// A struct for saving information about induction variables.
struct InductionInfo {
InductionInfo(Value *Start, InductionKind K) : StartValue(Start), IK(K) {}
InductionInfo() : StartValue(0), IK(IK_NoInduction) {}
/// pointer itself is an induction variable.
/// This check allows us to vectorize A[idx] into a wide load/store.
/// Returns:
- /// 0 - Stride is unknown or non consecutive.
+ /// 0 - Stride is unknown or non-consecutive.
/// 1 - Address is consecutive.
/// -1 - Address is consecutive, and decreasing.
int isConsecutivePtr(Value *Ptr);
unsigned getMaxSafeDepDistBytes() { return MaxSafeDepDistBytes; }
+ bool hasStride(Value *V) { return StrideSet.count(V); }
+ bool mustCheckStrides() { return !StrideSet.empty(); }
+ SmallPtrSet<Value *, 8>::iterator strides_begin() {
+ return StrideSet.begin();
+ }
+ SmallPtrSet<Value *, 8>::iterator strides_end() { return StrideSet.end(); }
+
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
/// if the PHI is not an induction variable.
InductionKind isInductionVariable(PHINode *Phi);
+ /// \brief Collect memory access with loop invariant strides.
+ ///
+ /// Looks for accesses like "a[i * StrideA]" where "StrideA" is loop
+ /// invariant.
+ void collectStridedAcccess(Value *LoadOrStoreInst);
+
/// The loop that we evaluate.
Loop *TheLoop;
/// Scev analysis.
ScalarEvolution *SE;
/// DataLayout analysis.
- DataLayout *DL;
+ const DataLayout *DL;
/// Dominators.
DominatorTree *DT;
/// Target Library Info.
bool HasFunNoNaNAttr;
unsigned MaxSafeDepDistBytes;
+
+ ValueToValueMap Strides;
+ SmallPtrSet<Value *, 8> StrideSet;
};
/// LoopVectorizationCostModel - estimates the expected speedups due to
LoopVectorizationCostModel(Loop *L, ScalarEvolution *SE, LoopInfo *LI,
LoopVectorizationLegality *Legal,
const TargetTransformInfo &TTI,
- DataLayout *DL, const TargetLibraryInfo *TLI)
+ const DataLayout *DL, const TargetLibraryInfo *TLI)
: TheLoop(L), SE(SE), LI(LI), Legal(Legal), TTI(TTI), DL(DL), TLI(TLI) {}
/// Information about vectorization costs
/// Vector target information.
const TargetTransformInfo &TTI;
/// Target data layout information.
- DataLayout *DL;
+ const DataLayout *DL;
/// Target Library Info.
const TargetLibraryInfo *TLI;
};
unsigned Width;
/// Vectorization unroll factor.
unsigned Unroll;
+ /// Vectorization forced (-1 not selected, 0 force disabled, 1 force enabled)
+ int Force;
- LoopVectorizeHints(const Loop *L)
+ LoopVectorizeHints(const Loop *L, bool DisableUnrolling)
: Width(VectorizationFactor)
- , Unroll(VectorizationUnroll)
+ , Unroll(DisableUnrolling ? 1 : VectorizationUnroll)
+ , Force(-1)
, LoopID(L->getLoopID()) {
getHints(L);
// The command line options override any loop metadata except for when
Width = VectorizationFactor;
if (VectorizationUnroll.getNumOccurrences() > 0)
Unroll = VectorizationUnroll;
+
+ DEBUG(if (DisableUnrolling && Unroll == 1)
+ dbgs() << "LV: Unrolling disabled by the pass manager\n");
}
/// Return the loop vectorizer metadata prefix.
Vals.push_back(LoopID->getOperand(i));
Vals.push_back(createHint(Context, Twine(Prefix(), "width").str(), Width));
+ Vals.push_back(createHint(Context, Twine(Prefix(), "unroll").str(), 1));
MDNode *NewLoopID = MDNode::get(Context, Vals);
// Set operand 0 to refer to the loop id itself.
unsigned Val = C->getZExtValue();
if (Hint == "width") {
- assert(isPowerOf2_32(Val) && Val <= MaxVectorWidth &&
- "Invalid width metadata");
- Width = Val;
+ if (isPowerOf2_32(Val) && Val <= MaxVectorWidth)
+ Width = Val;
+ else
+ DEBUG(dbgs() << "LV: ignoring invalid width hint metadata\n");
} else if (Hint == "unroll") {
- assert(isPowerOf2_32(Val) && Val <= MaxUnrollFactor &&
- "Invalid unroll metadata");
- Unroll = Val;
- } else
- DEBUG(dbgs() << "LV: ignoring unknown hint " << Hint);
+ if (isPowerOf2_32(Val) && Val <= MaxUnrollFactor)
+ Unroll = Val;
+ else
+ DEBUG(dbgs() << "LV: ignoring invalid unroll hint metadata\n");
+ } else if (Hint == "enable") {
+ if (C->getBitWidth() == 1)
+ Force = Val;
+ else
+ DEBUG(dbgs() << "LV: ignoring invalid enable hint metadata\n");
+ } else {
+ DEBUG(dbgs() << "LV: ignoring unknown hint " << Hint << '\n');
+ }
}
};
+static void addInnerLoop(Loop &L, SmallVectorImpl<Loop *> &V) {
+ if (L.empty())
+ return V.push_back(&L);
+
+ for (Loop *InnerL : L)
+ addInnerLoop(*InnerL, V);
+}
+
/// The LoopVectorize Pass.
-struct LoopVectorize : public LoopPass {
+struct LoopVectorize : public FunctionPass {
/// Pass identification, replacement for typeid
static char ID;
- explicit LoopVectorize() : LoopPass(ID) {
+ explicit LoopVectorize(bool NoUnrolling = false, bool AlwaysVectorize = true)
+ : FunctionPass(ID),
+ DisableUnrolling(NoUnrolling),
+ AlwaysVectorize(AlwaysVectorize) {
initializeLoopVectorizePass(*PassRegistry::getPassRegistry());
}
ScalarEvolution *SE;
- DataLayout *DL;
+ const DataLayout *DL;
LoopInfo *LI;
TargetTransformInfo *TTI;
DominatorTree *DT;
+ BlockFrequencyInfo *BFI;
TargetLibraryInfo *TLI;
+ bool DisableUnrolling;
+ bool AlwaysVectorize;
- virtual bool runOnLoop(Loop *L, LPPassManager &LPM) {
- // We only vectorize innermost loops.
- if (!L->empty())
- return false;
+ BlockFrequency ColdEntryFreq;
+ bool runOnFunction(Function &F) override {
SE = &getAnalysis<ScalarEvolution>();
- DL = getAnalysisIfAvailable<DataLayout>();
+ DataLayoutPass *DLP = getAnalysisIfAvailable<DataLayoutPass>();
+ DL = DLP ? &DLP->getDataLayout() : 0;
LI = &getAnalysis<LoopInfo>();
TTI = &getAnalysis<TargetTransformInfo>();
- DT = &getAnalysis<DominatorTree>();
+ DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree();
+ BFI = &getAnalysis<BlockFrequencyInfo>();
TLI = getAnalysisIfAvailable<TargetLibraryInfo>();
+ // Compute some weights outside of the loop over the loops. Compute this
+ // using a BranchProbability to re-use its scaling math.
+ const BranchProbability ColdProb(1, 5); // 20%
+ ColdEntryFreq = BlockFrequency(BFI->getEntryFreq()) * ColdProb;
+
+ // If the target claims to have no vector registers don't attempt
+ // vectorization.
+ if (!TTI->getNumberOfRegisters(true))
+ return false;
+
if (DL == NULL) {
- DEBUG(dbgs() << "LV: Not vectorizing because of missing data layout");
+ DEBUG(dbgs() << "\nLV: Not vectorizing " << F.getName()
+ << ": Missing data layout\n");
return false;
}
- DEBUG(dbgs() << "LV: Checking a loop in \"" <<
- L->getHeader()->getParent()->getName() << "\"\n");
+ // 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<Loop *, 8> Worklist;
+
+ for (Loop *L : *LI)
+ addInnerLoop(*L, Worklist);
+
+ // Now walk the identified inner loops.
+ bool Changed = false;
+ while (!Worklist.empty())
+ Changed |= processLoop(Worklist.pop_back_val());
+
+ // Process each loop nest in the function.
+ return Changed;
+ }
+
+ bool processLoop(Loop *L) {
+ assert(L->empty() && "Only process inner loops.");
+ DEBUG(dbgs() << "\nLV: Checking a loop in \""
+ << L->getHeader()->getParent()->getName() << "\" from "
+ << getDebugLocString(L->getHeader()->getFirstNonPHIOrDbg())
+ << "\n");
+
+ LoopVectorizeHints Hints(L, DisableUnrolling);
- LoopVectorizeHints Hints(L);
+ DEBUG(dbgs() << "LV: Loop hints:"
+ << " force=" << (Hints.Force == 0
+ ? "disabled"
+ : (Hints.Force == 1 ? "enabled" : "?"))
+ << " width=" << Hints.Width << " unroll=" << Hints.Unroll
+ << "\n");
- if (Hints.Width == 1) {
- DEBUG(dbgs() << "LV: Not vectorizing.\n");
+ if (Hints.Force == 0) {
+ DEBUG(dbgs() << "LV: Not vectorizing: #pragma vectorize disable.\n");
+ return false;
+ }
+
+ if (!AlwaysVectorize && Hints.Force != 1) {
+ DEBUG(dbgs() << "LV: Not vectorizing: No #pragma vectorize enable.\n");
+ return false;
+ }
+
+ if (Hints.Width == 1 && Hints.Unroll == 1) {
+ DEBUG(dbgs() << "LV: Not vectorizing: Disabled/already vectorized.\n");
return false;
}
// Check if it is legal to vectorize the loop.
LoopVectorizationLegality LVL(L, SE, DL, DT, TLI);
if (!LVL.canVectorize()) {
- DEBUG(dbgs() << "LV: Not vectorizing.\n");
+ DEBUG(dbgs() << "LV: Not vectorizing: Cannot prove legality.\n");
return false;
}
// Check the function attributes to find out if this function should be
// optimized for size.
Function *F = L->getHeader()->getParent();
- Attribute::AttrKind SzAttr = Attribute::OptimizeForSize;
- Attribute::AttrKind FlAttr = Attribute::NoImplicitFloat;
- unsigned FnIndex = AttributeSet::FunctionIndex;
- bool OptForSize = F->getAttributes().hasAttribute(FnIndex, SzAttr);
- bool NoFloat = F->getAttributes().hasAttribute(FnIndex, FlAttr);
+ bool OptForSize =
+ Hints.Force != 1 && F->hasFnAttribute(Attribute::OptimizeForSize);
+
+ // Compute the weighted frequency of this loop being executed and see if it
+ // is less than 20% of the function entry baseline frequency. Note that we
+ // always have a canonical loop here because we think we *can* vectoriez.
+ // FIXME: This is hidden behind a flag due to pervasive problems with
+ // exactly what block frequency models.
+ if (LoopVectorizeWithBlockFrequency) {
+ BlockFrequency LoopEntryFreq = BFI->getBlockFreq(L->getLoopPreheader());
+ if (Hints.Force != 1 && LoopEntryFreq < ColdEntryFreq)
+ OptForSize = true;
+ }
- if (NoFloat) {
+ // Check the function attributes to see if implicit floats are allowed.a
+ // FIXME: This check doesn't seem possibly correct -- what if the loop is
+ // an integer loop and the vector instructions selected are purely integer
+ // vector instructions?
+ if (F->hasFnAttribute(Attribute::NoImplicitFloat)) {
DEBUG(dbgs() << "LV: Can't vectorize when the NoImplicitFloat"
"attribute is used.\n");
return false;
}
// Select the optimal vectorization factor.
- LoopVectorizationCostModel::VectorizationFactor VF;
- VF = CM.selectVectorizationFactor(OptForSize, Hints.Width);
+ const LoopVectorizationCostModel::VectorizationFactor VF =
+ CM.selectVectorizationFactor(OptForSize, Hints.Width);
// Select the unroll factor.
- unsigned UF = CM.selectUnrollFactor(OptForSize, Hints.Unroll, VF.Width,
+ const unsigned UF = CM.selectUnrollFactor(OptForSize, Hints.Unroll, VF.Width,
VF.Cost);
+ DEBUG(dbgs() << "LV: Found a vectorizable loop ("
+ << VF.Width << ") in "
+ << getDebugLocString(L->getHeader()->getFirstNonPHIOrDbg())
+ << '\n');
+ DEBUG(dbgs() << "LV: Unroll Factor is " << UF << '\n');
+
if (VF.Width == 1) {
DEBUG(dbgs() << "LV: Vectorization is possible but not beneficial.\n");
- return false;
+ if (UF == 1)
+ return false;
+ DEBUG(dbgs() << "LV: Trying to at least unroll the loops.\n");
+ // We decided not to vectorize, but we may want to unroll.
+ InnerLoopUnroller Unroller(L, SE, LI, DT, DL, TLI, UF);
+ Unroller.vectorize(&LVL);
+ } else {
+ // If we decided that it is *legal* to vectorize the loop then do it.
+ InnerLoopVectorizer LB(L, SE, LI, DT, DL, TLI, VF.Width, UF);
+ LB.vectorize(&LVL);
}
- DEBUG(dbgs() << "LV: Found a vectorizable loop ("<< VF.Width << ") in "<<
- F->getParent()->getModuleIdentifier()<<"\n");
- DEBUG(dbgs() << "LV: Unroll Factor is " << UF << "\n");
-
- // If we decided that it is *legal* to vectorize the loop then do it.
- InnerLoopVectorizer LB(L, SE, LI, DT, DL, TLI, VF.Width, UF);
- LB.vectorize(&LVL);
-
// Mark the loop as already vectorized to avoid vectorizing again.
Hints.setAlreadyVectorized(L);
return true;
}
- virtual void getAnalysisUsage(AnalysisUsage &AU) const {
- LoopPass::getAnalysisUsage(AU);
+ void getAnalysisUsage(AnalysisUsage &AU) const override {
AU.addRequiredID(LoopSimplifyID);
AU.addRequiredID(LCSSAID);
- AU.addRequired<DominatorTree>();
+ AU.addRequired<BlockFrequencyInfo>();
+ AU.addRequired<DominatorTreeWrapperPass>();
AU.addRequired<LoopInfo>();
AU.addRequired<ScalarEvolution>();
AU.addRequired<TargetTransformInfo>();
AU.addPreserved<LoopInfo>();
- AU.addPreserved<DominatorTree>();
+ AU.addPreserved<DominatorTreeWrapperPass>();
}
};
// LoopVectorizationCostModel.
//===----------------------------------------------------------------------===//
-void
-LoopVectorizationLegality::RuntimePointerCheck::insert(ScalarEvolution *SE,
- Loop *Lp, Value *Ptr,
- bool WritePtr,
- unsigned DepSetId) {
- const SCEV *Sc = SE->getSCEV(Ptr);
+static Value *stripIntegerCast(Value *V) {
+ if (CastInst *CI = dyn_cast<CastInst>(V))
+ if (CI->getOperand(0)->getType()->isIntegerTy())
+ return CI->getOperand(0);
+ return V;
+}
+
+///\brief Replaces the symbolic stride in a pointer SCEV expression by one.
+///
+/// If \p OrigPtr is not null, use it to look up the stride value instead of
+/// \p Ptr.
+static const SCEV *replaceSymbolicStrideSCEV(ScalarEvolution *SE,
+ ValueToValueMap &PtrToStride,
+ Value *Ptr, Value *OrigPtr = 0) {
+
+ const SCEV *OrigSCEV = SE->getSCEV(Ptr);
+
+ // If there is an entry in the map return the SCEV of the pointer with the
+ // symbolic stride replaced by one.
+ ValueToValueMap::iterator SI = PtrToStride.find(OrigPtr ? OrigPtr : Ptr);
+ if (SI != PtrToStride.end()) {
+ Value *StrideVal = SI->second;
+
+ // Strip casts.
+ StrideVal = stripIntegerCast(StrideVal);
+
+ // Replace symbolic stride by one.
+ Value *One = ConstantInt::get(StrideVal->getType(), 1);
+ ValueToValueMap RewriteMap;
+ RewriteMap[StrideVal] = One;
+
+ const SCEV *ByOne =
+ SCEVParameterRewriter::rewrite(OrigSCEV, *SE, RewriteMap, true);
+ DEBUG(dbgs() << "LV: Replacing SCEV: " << *OrigSCEV << " by: " << *ByOne
+ << "\n");
+ return ByOne;
+ }
+
+ // Otherwise, just return the SCEV of the original pointer.
+ return SE->getSCEV(Ptr);
+}
+
+void LoopVectorizationLegality::RuntimePointerCheck::insert(
+ ScalarEvolution *SE, Loop *Lp, Value *Ptr, bool WritePtr, unsigned DepSetId,
+ ValueToValueMap &Strides) {
+ // Get the stride replaced scev.
+ const SCEV *Sc = replaceSymbolicStrideSCEV(SE, Strides, Ptr);
const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(Sc);
assert(AR && "Invalid addrec expression");
const SCEV *Ex = SE->getBackedgeTakenCount(Lp);
}
Value *InnerLoopVectorizer::getBroadcastInstrs(Value *V) {
- // Save the current insertion location.
- Instruction *Loc = Builder.GetInsertPoint();
-
// We need to place the broadcast of invariant variables outside the loop.
Instruction *Instr = dyn_cast<Instruction>(V);
- bool NewInstr = (Instr && Instr->getParent() == LoopVectorBody);
+ bool NewInstr =
+ (Instr && std::find(LoopVectorBody.begin(), LoopVectorBody.end(),
+ Instr->getParent()) != LoopVectorBody.end());
bool Invariant = OrigLoop->isLoopInvariant(V) && !NewInstr;
// Place the code for broadcasting invariant variables in the new preheader.
+ IRBuilder<>::InsertPointGuard Guard(Builder);
if (Invariant)
Builder.SetInsertPoint(LoopVectorPreHeader->getTerminator());
// Broadcast the scalar into all locations in the vector.
Value *Shuf = Builder.CreateVectorSplat(VF, V, "broadcast");
- // Restore the builder insertion point.
- if (Invariant)
- Builder.SetInsertPoint(Loc);
-
return Shuf;
}
return Builder.CreateAdd(Val, Cv, "induction");
}
+/// \brief Find the operand of the GEP that should be checked for consecutive
+/// stores. This ignores trailing indices that have no effect on the final
+/// pointer.
+static unsigned getGEPInductionOperand(const DataLayout *DL,
+ const GetElementPtrInst *Gep) {
+ unsigned LastOperand = Gep->getNumOperands() - 1;
+ unsigned GEPAllocSize = DL->getTypeAllocSize(
+ cast<PointerType>(Gep->getType()->getScalarType())->getElementType());
+
+ // Walk backwards and try to peel off zeros.
+ while (LastOperand > 1 && match(Gep->getOperand(LastOperand), m_Zero())) {
+ // Find the type we're currently indexing into.
+ gep_type_iterator GEPTI = gep_type_begin(Gep);
+ std::advance(GEPTI, LastOperand - 1);
+
+ // If it's a type with the same allocation size as the result of the GEP we
+ // can peel off the zero index.
+ if (DL->getTypeAllocSize(*GEPTI) != GEPAllocSize)
+ break;
+ --LastOperand;
+ }
+
+ return LastOperand;
+}
+
int LoopVectorizationLegality::isConsecutivePtr(Value *Ptr) {
- assert(Ptr->getType()->isPointerTy() && "Unexpected non ptr");
+ assert(Ptr->getType()->isPointerTy() && "Unexpected non-ptr");
// Make sure that the pointer does not point to structs.
- if (cast<PointerType>(Ptr->getType())->getElementType()->isAggregateType())
+ if (Ptr->getType()->getPointerElementType()->isAggregateType())
return 0;
// If this value is a pointer induction variable we know it is consecutive.
return 0;
unsigned NumOperands = Gep->getNumOperands();
- Value *LastIndex = Gep->getOperand(NumOperands - 1);
-
Value *GpPtr = Gep->getPointerOperand();
// If this GEP value is a consecutive pointer induction variable and all of
// the indices are constant then we know it is consecutive. We can
return -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))
+ unsigned InductionOperand = getGEPInductionOperand(DL, Gep);
+
+ // Check that all of the gep indices are uniform except for our induction
+ // operand.
+ for (unsigned i = 0; i != NumOperands; ++i)
+ if (i != InductionOperand &&
+ !SE->isLoopInvariant(SE->getSCEV(Gep->getOperand(i)), TheLoop))
return 0;
- // We can emit wide load/stores only if the last index is the induction
- // variable.
- const SCEV *Last = SE->getSCEV(LastIndex);
+ // We can emit wide load/stores only if the last non-zero index is the
+ // induction variable.
+ const SCEV *Last = 0;
+ if (!Strides.count(Gep))
+ Last = SE->getSCEV(Gep->getOperand(InductionOperand));
+ else {
+ // Because of the multiplication by a stride we can have a s/zext cast.
+ // We are going to replace this stride by 1 so the cast is safe to ignore.
+ //
+ // %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
+ // %0 = trunc i64 %indvars.iv to i32
+ // %mul = mul i32 %0, %Stride1
+ // %idxprom = zext i32 %mul to i64 << Safe cast.
+ // %arrayidx = getelementptr inbounds i32* %B, i64 %idxprom
+ //
+ Last = replaceSymbolicStrideSCEV(SE, Strides,
+ Gep->getOperand(InductionOperand), Gep);
+ if (const SCEVCastExpr *C = dyn_cast<SCEVCastExpr>(Last))
+ Last =
+ (C->getSCEVType() == scSignExtend || C->getSCEVType() == scZeroExtend)
+ ? C->getOperand()
+ : Last;
+ }
if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(Last)) {
const SCEV *Step = AR->getStepRecurrence(*SE);
assert(V != Induction && "The new induction variable should not be used.");
assert(!V->getType()->isVectorTy() && "Can't widen a vector");
+ // If we have a stride that is replaced by one, do it here.
+ if (Legal->hasStride(V))
+ V = ConstantInt::get(V->getType(), 1);
+
// If we have this scalar in the map, return it.
if (WidenMap.has(V))
return WidenMap.get(V);
"reverse");
}
-
-void InnerLoopVectorizer::vectorizeMemoryInstruction(Instruction *Instr,
- LoopVectorizationLegality *Legal) {
+void InnerLoopVectorizer::vectorizeMemoryInstruction(Instruction *Instr) {
// Attempt to issue a wide load.
LoadInst *LI = dyn_cast<LoadInst>(Instr);
StoreInst *SI = dyn_cast<StoreInst>(Instr);
Type *DataTy = VectorType::get(ScalarDataTy, VF);
Value *Ptr = LI ? LI->getPointerOperand() : SI->getPointerOperand();
unsigned Alignment = LI ? LI->getAlignment() : SI->getAlignment();
+ // An alignment of 0 means target abi alignment. We need to use the scalar's
+ // target abi alignment in such a case.
+ if (!Alignment)
+ Alignment = DL->getABITypeAlignment(ScalarDataTy);
unsigned AddressSpace = Ptr->getType()->getPointerAddressSpace();
unsigned ScalarAllocatedSize = DL->getTypeAllocSize(ScalarDataTy);
unsigned VectorElementSize = DL->getTypeStoreSize(DataTy)/VF;
+ if (SI && Legal->blockNeedsPredication(SI->getParent()))
+ return scalarizeInstruction(Instr, true);
+
if (ScalarAllocatedSize != VectorElementSize)
return scalarizeInstruction(Instr);
- // If the pointer is loop invariant or if it is non consecutive,
+ // If the pointer is loop invariant or if it is non-consecutive,
// scalarize the load.
int ConsecutiveStride = Legal->isConsecutivePtr(Ptr);
bool Reverse = ConsecutiveStride < 0;
// 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();
- unsigned LastOperand = NumOperands - 1;
+ unsigned InductionOperand = getGEPInductionOperand(DL, Gep);
// Create the new GEP with the new induction variable.
GetElementPtrInst *Gep2 = cast<GetElementPtrInst>(Gep->clone());
Instruction *GepOperandInst = dyn_cast<Instruction>(GepOperand);
// Update last index or loop invariant instruction anchored in loop.
- if (i == LastOperand ||
+ if (i == InductionOperand ||
(GepOperandInst && OrigLoop->contains(GepOperandInst))) {
- assert((i == LastOperand ||
+ assert((i == InductionOperand ||
SE->isLoopInvariant(SE->getSCEV(GepOperandInst), OrigLoop)) &&
"Must be last index or loop invariant");
PartPtr = Builder.CreateGEP(PartPtr, Builder.getInt32(1 - VF));
}
- Value *VecPtr = Builder.CreateBitCast(PartPtr, DataTy->getPointerTo(AddressSpace));
+ Value *VecPtr = Builder.CreateBitCast(PartPtr,
+ DataTy->getPointerTo(AddressSpace));
Builder.CreateStore(StoredVal[Part], VecPtr)->setAlignment(Alignment);
}
return;
PartPtr = Builder.CreateGEP(PartPtr, Builder.getInt32(1 - VF));
}
- Value *VecPtr = Builder.CreateBitCast(PartPtr, DataTy->getPointerTo(AddressSpace));
+ Value *VecPtr = Builder.CreateBitCast(PartPtr,
+ DataTy->getPointerTo(AddressSpace));
Value *LI = Builder.CreateLoad(VecPtr, "wide.load");
cast<LoadInst>(LI)->setAlignment(Alignment);
Entry[Part] = Reverse ? reverseVector(LI) : LI;
}
}
-void InnerLoopVectorizer::scalarizeInstruction(Instruction *Instr) {
+void InnerLoopVectorizer::scalarizeInstruction(Instruction *Instr, bool IfPredicateStore) {
assert(!Instr->getType()->isAggregateType() && "Can't handle vectors");
// Holds vector parameters or scalars, in case of uniform vals.
SmallVector<VectorParts, 4> Params;
// Create a new entry in the WidenMap and initialize it to Undef or Null.
VectorParts &VecResults = WidenMap.splat(Instr, UndefVec);
+ Instruction *InsertPt = Builder.GetInsertPoint();
+ BasicBlock *IfBlock = Builder.GetInsertBlock();
+ BasicBlock *CondBlock = 0;
+
+ VectorParts Cond;
+ Loop *VectorLp = 0;
+ if (IfPredicateStore) {
+ assert(Instr->getParent()->getSinglePredecessor() &&
+ "Only support single predecessor blocks");
+ Cond = createEdgeMask(Instr->getParent()->getSinglePredecessor(),
+ Instr->getParent());
+ VectorLp = LI->getLoopFor(IfBlock);
+ assert(VectorLp && "Must have a loop for this block");
+ }
+
// For each vector unroll 'part':
for (unsigned Part = 0; Part < UF; ++Part) {
// For each scalar that we create:
for (unsigned Width = 0; Width < VF; ++Width) {
+
+ // Start if-block.
+ Value *Cmp = 0;
+ if (IfPredicateStore) {
+ Cmp = Builder.CreateExtractElement(Cond[Part], Builder.getInt32(Width));
+ Cmp = Builder.CreateICmp(ICmpInst::ICMP_EQ, Cmp, ConstantInt::get(Cmp->getType(), 1));
+ CondBlock = IfBlock->splitBasicBlock(InsertPt, "cond.store");
+ LoopVectorBody.push_back(CondBlock);
+ VectorLp->addBasicBlockToLoop(CondBlock, LI->getBase());
+ // Update Builder with newly created basic block.
+ Builder.SetInsertPoint(InsertPt);
+ }
+
Instruction *Cloned = Instr->clone();
if (!IsVoidRetTy)
Cloned->setName(Instr->getName() + ".cloned");
- // Replace the operands of the cloned instrucions with extracted scalars.
+ // Replace the operands of the cloned instructions with extracted scalars.
for (unsigned op = 0, e = Instr->getNumOperands(); op != e; ++op) {
Value *Op = Params[op][Part];
// Param is a vector. Need to extract the right lane.
if (!IsVoidRetTy)
VecResults[Part] = Builder.CreateInsertElement(VecResults[Part], Cloned,
Builder.getInt32(Width));
+ // End if-block.
+ if (IfPredicateStore) {
+ BasicBlock *NewIfBlock = CondBlock->splitBasicBlock(InsertPt, "else");
+ LoopVectorBody.push_back(NewIfBlock);
+ VectorLp->addBasicBlockToLoop(NewIfBlock, LI->getBase());
+ Builder.SetInsertPoint(InsertPt);
+ Instruction *OldBr = IfBlock->getTerminator();
+ BranchInst::Create(CondBlock, NewIfBlock, Cmp, OldBr);
+ OldBr->eraseFromParent();
+ IfBlock = NewIfBlock;
+ }
}
}
}
-Instruction *
-InnerLoopVectorizer::addRuntimeCheck(LoopVectorizationLegality *Legal,
- Instruction *Loc) {
+static Instruction *getFirstInst(Instruction *FirstInst, Value *V,
+ Instruction *Loc) {
+ if (FirstInst)
+ return FirstInst;
+ if (Instruction *I = dyn_cast<Instruction>(V))
+ return I->getParent() == Loc->getParent() ? I : 0;
+ return 0;
+}
+
+std::pair<Instruction *, Instruction *>
+InnerLoopVectorizer::addStrideCheck(Instruction *Loc) {
+ Instruction *tnullptr = 0;
+ if (!Legal->mustCheckStrides())
+ return std::pair<Instruction *, Instruction *>(tnullptr, tnullptr);
+
+ IRBuilder<> ChkBuilder(Loc);
+
+ // Emit checks.
+ Value *Check = 0;
+ Instruction *FirstInst = 0;
+ for (SmallPtrSet<Value *, 8>::iterator SI = Legal->strides_begin(),
+ SE = Legal->strides_end();
+ SI != SE; ++SI) {
+ Value *Ptr = stripIntegerCast(*SI);
+ Value *C = ChkBuilder.CreateICmpNE(Ptr, ConstantInt::get(Ptr->getType(), 1),
+ "stride.chk");
+ // Store the first instruction we create.
+ FirstInst = getFirstInst(FirstInst, C, Loc);
+ if (Check)
+ Check = ChkBuilder.CreateOr(Check, C);
+ else
+ Check = C;
+ }
+
+ // We have to do this trickery because the IRBuilder might fold the check to a
+ // constant expression in which case there is no Instruction anchored in a
+ // the block.
+ LLVMContext &Ctx = Loc->getContext();
+ Instruction *TheCheck =
+ BinaryOperator::CreateAnd(Check, ConstantInt::getTrue(Ctx));
+ ChkBuilder.Insert(TheCheck, "stride.not.one");
+ FirstInst = getFirstInst(FirstInst, TheCheck, Loc);
+
+ return std::make_pair(FirstInst, TheCheck);
+}
+
+std::pair<Instruction *, Instruction *>
+InnerLoopVectorizer::addRuntimeCheck(Instruction *Loc) {
LoopVectorizationLegality::RuntimePointerCheck *PtrRtCheck =
Legal->getRuntimePointerCheck();
+ Instruction *tnullptr = 0;
if (!PtrRtCheck->Need)
- return NULL;
+ return std::pair<Instruction *, Instruction *>(tnullptr, tnullptr);
unsigned NumPointers = PtrRtCheck->Pointers.size();
SmallVector<TrackingVH<Value> , 2> Starts;
SmallVector<TrackingVH<Value> , 2> Ends;
+ LLVMContext &Ctx = Loc->getContext();
SCEVExpander Exp(*SE, "induction");
-
- // Use this type for pointer arithmetic.
- Type* PtrArithTy = Type::getInt8PtrTy(Loc->getContext(), 0);
+ Instruction *FirstInst = 0;
for (unsigned i = 0; i < NumPointers; ++i) {
Value *Ptr = PtrRtCheck->Pointers[i];
Starts.push_back(Ptr);
Ends.push_back(Ptr);
} else {
- DEBUG(dbgs() << "LV: Adding RT check for range:" << *Ptr <<"\n");
+ DEBUG(dbgs() << "LV: Adding RT check for range:" << *Ptr << '\n');
+ unsigned AS = Ptr->getType()->getPointerAddressSpace();
+
+ // Use this type for pointer arithmetic.
+ Type *PtrArithTy = Type::getInt8PtrTy(Ctx, AS);
Value *Start = Exp.expandCodeFor(PtrRtCheck->Starts[i], PtrArithTy, Loc);
Value *End = Exp.expandCodeFor(PtrRtCheck->Ends[i], PtrArithTy, Loc);
if (PtrRtCheck->DependencySetId[i] == PtrRtCheck->DependencySetId[j])
continue;
- Value *Start0 = ChkBuilder.CreateBitCast(Starts[i], PtrArithTy, "bc");
- Value *Start1 = ChkBuilder.CreateBitCast(Starts[j], PtrArithTy, "bc");
- Value *End0 = ChkBuilder.CreateBitCast(Ends[i], PtrArithTy, "bc");
- Value *End1 = ChkBuilder.CreateBitCast(Ends[j], PtrArithTy, "bc");
+ unsigned AS0 = Starts[i]->getType()->getPointerAddressSpace();
+ unsigned AS1 = Starts[j]->getType()->getPointerAddressSpace();
+
+ assert((AS0 == Ends[j]->getType()->getPointerAddressSpace()) &&
+ (AS1 == Ends[i]->getType()->getPointerAddressSpace()) &&
+ "Trying to bounds check pointers with different address spaces");
+
+ Type *PtrArithTy0 = Type::getInt8PtrTy(Ctx, AS0);
+ Type *PtrArithTy1 = Type::getInt8PtrTy(Ctx, AS1);
+
+ Value *Start0 = ChkBuilder.CreateBitCast(Starts[i], PtrArithTy0, "bc");
+ Value *Start1 = ChkBuilder.CreateBitCast(Starts[j], PtrArithTy1, "bc");
+ Value *End0 = ChkBuilder.CreateBitCast(Ends[i], PtrArithTy1, "bc");
+ Value *End1 = ChkBuilder.CreateBitCast(Ends[j], PtrArithTy0, "bc");
Value *Cmp0 = ChkBuilder.CreateICmpULE(Start0, End1, "bound0");
+ FirstInst = getFirstInst(FirstInst, Cmp0, Loc);
Value *Cmp1 = ChkBuilder.CreateICmpULE(Start1, End0, "bound1");
+ FirstInst = getFirstInst(FirstInst, Cmp1, Loc);
Value *IsConflict = ChkBuilder.CreateAnd(Cmp0, Cmp1, "found.conflict");
- if (MemoryRuntimeCheck)
+ FirstInst = getFirstInst(FirstInst, IsConflict, Loc);
+ if (MemoryRuntimeCheck) {
IsConflict = ChkBuilder.CreateOr(MemoryRuntimeCheck, IsConflict,
"conflict.rdx");
+ FirstInst = getFirstInst(FirstInst, IsConflict, Loc);
+ }
MemoryRuntimeCheck = IsConflict;
}
}
// We have to do this trickery because the IRBuilder might fold the check to a
// constant expression in which case there is no Instruction anchored in a
// the block.
- LLVMContext &Ctx = Loc->getContext();
- Instruction * Check = BinaryOperator::CreateAnd(MemoryRuntimeCheck,
- ConstantInt::getTrue(Ctx));
+ Instruction *Check = BinaryOperator::CreateAnd(MemoryRuntimeCheck,
+ ConstantInt::getTrue(Ctx));
ChkBuilder.Insert(Check, "memcheck.conflict");
- return Check;
+ FirstInst = getFirstInst(FirstInst, Check, Loc);
+ return std::make_pair(FirstInst, Check);
}
-void
-InnerLoopVectorizer::createEmptyLoop(LoopVectorizationLegality *Legal) {
+void InnerLoopVectorizer::createEmptyLoop() {
/*
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
const SCEV *ExitCount = SE->getBackedgeTakenCount(OrigLoop);
assert(ExitCount != SE->getCouldNotCompute() && "Invalid loop count");
+ // The exit count might have the type of i64 while the phi is i32. This can
+ // happen if we have an induction variable that is sign extended before the
+ // compare. The only way that we get a backedge taken count is that the
+ // induction variable was signed and as such will not overflow. In such a case
+ // truncation is legal.
+ if (ExitCount->getType()->getPrimitiveSizeInBits() >
+ IdxTy->getPrimitiveSizeInBits())
+ ExitCount = SE->getTruncateOrNoop(ExitCount, IdxTy);
+
+ ExitCount = SE->getNoopOrZeroExtend(ExitCount, IdxTy);
// Get the total trip count from the count by adding 1.
ExitCount = SE->getAddExpr(ExitCount,
SE->getConstant(ExitCount->getType(), 1));
// Use this IR builder to create the loop instructions (Phi, Br, Cmp)
// inside the loop.
- Builder.SetInsertPoint(VecBody->getFirstInsertionPt());
+ Builder.SetInsertPoint(VecBody->getFirstNonPHI());
// Generate the induction variable.
setDebugLocFromInst(Builder, getDebugLocFromInstOrOperands(OldInduction));
BasicBlock *LastBypassBlock = BypassBlock;
+ // Generate the code to check that the strides we assumed to be one are really
+ // one. We want the new basic block to start at the first instruction in a
+ // sequence of instructions that form a check.
+ Instruction *StrideCheck;
+ Instruction *FirstCheckInst;
+ std::tie(FirstCheckInst, StrideCheck) =
+ addStrideCheck(BypassBlock->getTerminator());
+ if (StrideCheck) {
+ // Create a new block containing the stride check.
+ BasicBlock *CheckBlock =
+ BypassBlock->splitBasicBlock(FirstCheckInst, "vector.stridecheck");
+ if (ParentLoop)
+ ParentLoop->addBasicBlockToLoop(CheckBlock, LI->getBase());
+ LoopBypassBlocks.push_back(CheckBlock);
+
+ // Replace the branch into the memory check block with a conditional branch
+ // for the "few elements case".
+ Instruction *OldTerm = BypassBlock->getTerminator();
+ BranchInst::Create(MiddleBlock, CheckBlock, Cmp, OldTerm);
+ OldTerm->eraseFromParent();
+
+ Cmp = StrideCheck;
+ LastBypassBlock = CheckBlock;
+ }
+
// Generate the code that checks in runtime if arrays overlap. We put the
// checks into a separate block to make the more common case of few elements
// faster.
- Instruction *MemRuntimeCheck = addRuntimeCheck(Legal,
- BypassBlock->getTerminator());
+ Instruction *MemRuntimeCheck;
+ std::tie(FirstCheckInst, MemRuntimeCheck) =
+ addRuntimeCheck(LastBypassBlock->getTerminator());
if (MemRuntimeCheck) {
// Create a new block containing the memory check.
- BasicBlock *CheckBlock = BypassBlock->splitBasicBlock(MemRuntimeCheck,
- "vector.memcheck");
+ BasicBlock *CheckBlock =
+ LastBypassBlock->splitBasicBlock(MemRuntimeCheck, "vector.memcheck");
if (ParentLoop)
ParentLoop->addBasicBlockToLoop(CheckBlock, LI->getBase());
LoopBypassBlocks.push_back(CheckBlock);
// Replace the branch into the memory check block with a conditional branch
// for the "few elements case".
- Instruction *OldTerm = BypassBlock->getTerminator();
+ Instruction *OldTerm = LastBypassBlock->getTerminator();
BranchInst::Create(MiddleBlock, CheckBlock, Cmp, OldTerm);
OldTerm->eraseFromParent();
LoopScalarPreHeader = ScalarPH;
LoopMiddleBlock = MiddleBlock;
LoopExitBlock = ExitBlock;
- LoopVectorBody = VecBody;
+ LoopVectorBody.push_back(VecBody);
LoopScalarBody = OldBasicBlock;
+
+ LoopVectorizeHints Hints(Lp, true);
+ Hints.setAlreadyVectorized(Lp);
}
/// This function returns the identity element (or neutral element) for
}
}
+static Intrinsic::ID checkUnaryFloatSignature(const CallInst &I,
+ Intrinsic::ID ValidIntrinsicID) {
+ if (I.getNumArgOperands() != 1 ||
+ !I.getArgOperand(0)->getType()->isFloatingPointTy() ||
+ I.getType() != I.getArgOperand(0)->getType() ||
+ !I.onlyReadsMemory())
+ return Intrinsic::not_intrinsic;
+
+ return ValidIntrinsicID;
+}
+
+static Intrinsic::ID checkBinaryFloatSignature(const CallInst &I,
+ Intrinsic::ID ValidIntrinsicID) {
+ if (I.getNumArgOperands() != 2 ||
+ !I.getArgOperand(0)->getType()->isFloatingPointTy() ||
+ !I.getArgOperand(1)->getType()->isFloatingPointTy() ||
+ I.getType() != I.getArgOperand(0)->getType() ||
+ I.getType() != I.getArgOperand(1)->getType() ||
+ !I.onlyReadsMemory())
+ return Intrinsic::not_intrinsic;
+
+ return ValidIntrinsicID;
+}
+
+
static Intrinsic::ID
getIntrinsicIDForCall(CallInst *CI, const TargetLibraryInfo *TLI) {
// If we have an intrinsic call, check if it is trivially vectorizable.
if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(CI)) {
- switch (II->getIntrinsicID()) {
- case Intrinsic::sqrt:
- case Intrinsic::sin:
- case Intrinsic::cos:
- case Intrinsic::exp:
- case Intrinsic::exp2:
- case Intrinsic::log:
- case Intrinsic::log10:
- case Intrinsic::log2:
- case Intrinsic::fabs:
- case Intrinsic::floor:
- case Intrinsic::ceil:
- case Intrinsic::trunc:
- case Intrinsic::rint:
- case Intrinsic::nearbyint:
- case Intrinsic::pow:
- case Intrinsic::fma:
- case Intrinsic::fmuladd:
- return II->getIntrinsicID();
- default:
+ Intrinsic::ID ID = II->getIntrinsicID();
+ if (isTriviallyVectorizable(ID) || ID == Intrinsic::lifetime_start ||
+ ID == Intrinsic::lifetime_end)
+ return ID;
+ else
return Intrinsic::not_intrinsic;
- }
}
if (!TLI)
LibFunc::Func Func;
Function *F = CI->getCalledFunction();
// We're going to make assumptions on the semantics of the functions, check
- // that the target knows that it's available in this environment.
- if (!F || !TLI->getLibFunc(F->getName(), Func))
+ // that the target knows that it's available in this environment and it does
+ // not have local linkage.
+ if (!F || F->hasLocalLinkage() || !TLI->getLibFunc(F->getName(), Func))
return Intrinsic::not_intrinsic;
// Otherwise check if we have a call to a function that can be turned into a
case LibFunc::sin:
case LibFunc::sinf:
case LibFunc::sinl:
- return Intrinsic::sin;
+ return checkUnaryFloatSignature(*CI, Intrinsic::sin);
case LibFunc::cos:
case LibFunc::cosf:
case LibFunc::cosl:
- return Intrinsic::cos;
+ return checkUnaryFloatSignature(*CI, Intrinsic::cos);
case LibFunc::exp:
case LibFunc::expf:
case LibFunc::expl:
- return Intrinsic::exp;
+ return checkUnaryFloatSignature(*CI, Intrinsic::exp);
case LibFunc::exp2:
case LibFunc::exp2f:
case LibFunc::exp2l:
- return Intrinsic::exp2;
+ return checkUnaryFloatSignature(*CI, Intrinsic::exp2);
case LibFunc::log:
case LibFunc::logf:
case LibFunc::logl:
- return Intrinsic::log;
+ return checkUnaryFloatSignature(*CI, Intrinsic::log);
case LibFunc::log10:
case LibFunc::log10f:
case LibFunc::log10l:
- return Intrinsic::log10;
+ return checkUnaryFloatSignature(*CI, Intrinsic::log10);
case LibFunc::log2:
case LibFunc::log2f:
case LibFunc::log2l:
- return Intrinsic::log2;
+ return checkUnaryFloatSignature(*CI, Intrinsic::log2);
case LibFunc::fabs:
case LibFunc::fabsf:
case LibFunc::fabsl:
- return Intrinsic::fabs;
+ return checkUnaryFloatSignature(*CI, Intrinsic::fabs);
+ case LibFunc::copysign:
+ case LibFunc::copysignf:
+ case LibFunc::copysignl:
+ return checkBinaryFloatSignature(*CI, Intrinsic::copysign);
case LibFunc::floor:
case LibFunc::floorf:
case LibFunc::floorl:
- return Intrinsic::floor;
+ return checkUnaryFloatSignature(*CI, Intrinsic::floor);
case LibFunc::ceil:
case LibFunc::ceilf:
case LibFunc::ceill:
- return Intrinsic::ceil;
+ return checkUnaryFloatSignature(*CI, Intrinsic::ceil);
case LibFunc::trunc:
case LibFunc::truncf:
case LibFunc::truncl:
- return Intrinsic::trunc;
+ return checkUnaryFloatSignature(*CI, Intrinsic::trunc);
case LibFunc::rint:
case LibFunc::rintf:
case LibFunc::rintl:
- return Intrinsic::rint;
+ return checkUnaryFloatSignature(*CI, Intrinsic::rint);
case LibFunc::nearbyint:
case LibFunc::nearbyintf:
case LibFunc::nearbyintl:
- return Intrinsic::nearbyint;
+ return checkUnaryFloatSignature(*CI, Intrinsic::nearbyint);
+ case LibFunc::round:
+ case LibFunc::roundf:
+ case LibFunc::roundl:
+ return checkUnaryFloatSignature(*CI, Intrinsic::round);
case LibFunc::pow:
case LibFunc::powf:
case LibFunc::powl:
- return Intrinsic::pow;
+ return checkBinaryFloatSignature(*CI, Intrinsic::pow);
}
return Intrinsic::not_intrinsic;
}
Value *Cmp;
- if (RK == LoopVectorizationLegality::MRK_FloatMin || RK == LoopVectorizationLegality::MRK_FloatMax)
+ if (RK == LoopVectorizationLegality::MRK_FloatMin ||
+ RK == LoopVectorizationLegality::MRK_FloatMax)
Cmp = Builder.CreateFCmp(P, Left, Right, "rdx.minmax.cmp");
else
Cmp = Builder.CreateICmp(P, Left, Right, "rdx.minmax.cmp");
return Select;
}
-void
-InnerLoopVectorizer::vectorizeLoop(LoopVectorizationLegality *Legal) {
+namespace {
+struct CSEDenseMapInfo {
+ static bool canHandle(Instruction *I) {
+ return isa<InsertElementInst>(I) || isa<ExtractElementInst>(I) ||
+ isa<ShuffleVectorInst>(I) || isa<GetElementPtrInst>(I);
+ }
+ static inline Instruction *getEmptyKey() {
+ return DenseMapInfo<Instruction *>::getEmptyKey();
+ }
+ static inline Instruction *getTombstoneKey() {
+ return DenseMapInfo<Instruction *>::getTombstoneKey();
+ }
+ static unsigned getHashValue(Instruction *I) {
+ assert(canHandle(I) && "Unknown instruction!");
+ return hash_combine(I->getOpcode(), hash_combine_range(I->value_op_begin(),
+ I->value_op_end()));
+ }
+ static bool isEqual(Instruction *LHS, Instruction *RHS) {
+ if (LHS == getEmptyKey() || RHS == getEmptyKey() ||
+ LHS == getTombstoneKey() || RHS == getTombstoneKey())
+ return LHS == RHS;
+ return LHS->isIdenticalTo(RHS);
+ }
+};
+}
+
+/// \brief Check whether this block is a predicated block.
+/// Due to if predication of stores we might create a sequence of "if(pred) a[i]
+/// = ...; " blocks. We start with one vectorized basic block. For every
+/// conditional block we split this vectorized block. Therefore, every second
+/// block will be a predicated one.
+static bool isPredicatedBlock(unsigned BlockNum) {
+ return BlockNum % 2;
+}
+
+///\brief Perform cse of induction variable instructions.
+static void cse(SmallVector<BasicBlock *, 4> &BBs) {
+ // Perform simple cse.
+ SmallDenseMap<Instruction *, Instruction *, 4, CSEDenseMapInfo> CSEMap;
+ for (unsigned i = 0, e = BBs.size(); i != e; ++i) {
+ BasicBlock *BB = BBs[i];
+ for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E;) {
+ Instruction *In = I++;
+
+ 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;
+ }
+ // Ignore instructions in conditional blocks. We create "if (pred) a[i] =
+ // ...;" blocks for predicated stores. Every second block is a predicated
+ // block.
+ if (isPredicatedBlock(i))
+ continue;
+
+ CSEMap[In] = In;
+ }
+ }
+}
+
+/// \brief Adds a 'fast' flag to floating point operations.
+static Value *addFastMathFlag(Value *V) {
+ if (isa<FPMathOperator>(V)){
+ FastMathFlags Flags;
+ Flags.setUnsafeAlgebra();
+ cast<Instruction>(V)->setFastMathFlags(Flags);
+ }
+ return V;
+}
+
+void InnerLoopVectorizer::vectorizeLoop() {
//===------------------------------------------------===//
//
// Notice: any optimization or new instruction that go
// Vectorize all of the blocks in the original loop.
for (LoopBlocksDFS::RPOIterator bb = DFS.beginRPO(),
be = DFS.endRPO(); bb != be; ++bb)
- vectorizeBlockInLoop(Legal, *bb, &RdxPHIsToFix);
+ vectorizeBlockInLoop(*bb, &RdxPHIsToFix);
// At this point every instruction in the original loop is widened to
// a vector form. We are almost done. Now, we need to fix the PHI nodes
setDebugLocFromInst(Builder, RdxDesc.StartValue);
// We need to generate a reduction vector from the incoming scalar.
- // To do so, we need to generate the 'identity' vector and overide
+ // To do so, we need to generate the 'identity' vector and override
// one of the elements with the incoming scalar reduction. We need
// to do it in the vector-loop preheader.
Builder.SetInsertPoint(LoopBypassBlocks.front()->getTerminator());
if (RdxDesc.Kind == LoopVectorizationLegality::RK_IntegerMinMax ||
RdxDesc.Kind == LoopVectorizationLegality::RK_FloatMinMax) {
// MinMax reduction have the start value as their identify.
- VectorStart = Identity = Builder.CreateVectorSplat(VF, RdxDesc.StartValue,
- "minmax.ident");
+ if (VF == 1) {
+ VectorStart = Identity = RdxDesc.StartValue;
+ } else {
+ VectorStart = Identity = Builder.CreateVectorSplat(VF,
+ RdxDesc.StartValue,
+ "minmax.ident");
+ }
} else {
+ // Handle other reduction kinds:
Constant *Iden =
- LoopVectorizationLegality::getReductionIdentity(RdxDesc.Kind,
- VecTy->getScalarType());
- Identity = ConstantVector::getSplat(VF, Iden);
-
- // This vector is the Identity vector where the first element is the
- // incoming scalar reduction.
- VectorStart = Builder.CreateInsertElement(Identity,
- RdxDesc.StartValue, Zero);
+ LoopVectorizationLegality::getReductionIdentity(RdxDesc.Kind,
+ VecTy->getScalarType());
+ if (VF == 1) {
+ Identity = Iden;
+ // This vector is the Identity vector where the first element is the
+ // incoming scalar reduction.
+ VectorStart = RdxDesc.StartValue;
+ } else {
+ Identity = ConstantVector::getSplat(VF, Iden);
+
+ // This vector is the Identity vector where the first element is the
+ // incoming scalar reduction.
+ VectorStart = Builder.CreateInsertElement(Identity,
+ RdxDesc.StartValue, Zero);
+ }
}
// Fix the vector-loop phi.
// first unroll part.
Value *StartVal = (part == 0) ? VectorStart : Identity;
cast<PHINode>(VecRdxPhi[part])->addIncoming(StartVal, VecPreheader);
- cast<PHINode>(VecRdxPhi[part])->addIncoming(Val[part], LoopVectorBody);
+ cast<PHINode>(VecRdxPhi[part])->addIncoming(Val[part],
+ LoopVectorBody.back());
}
// Before each round, move the insertion point right between
Value *StartVal = (part == 0) ? VectorStart : Identity;
for (unsigned I = 0, E = LoopBypassBlocks.size(); I != E; ++I)
NewPhi->addIncoming(StartVal, LoopBypassBlocks[I]);
- NewPhi->addIncoming(RdxExitVal[part], LoopVectorBody);
+ NewPhi->addIncoming(RdxExitVal[part],
+ LoopVectorBody.back());
RdxParts.push_back(NewPhi);
}
setDebugLocFromInst(Builder, ReducedPartRdx);
for (unsigned part = 1; part < UF; ++part) {
if (Op != Instruction::ICmp && Op != Instruction::FCmp)
- ReducedPartRdx = Builder.CreateBinOp((Instruction::BinaryOps)Op,
- RdxParts[part], ReducedPartRdx,
- "bin.rdx");
+ // Floating point operations had to be 'fast' to enable the reduction.
+ ReducedPartRdx = addFastMathFlag(
+ Builder.CreateBinOp((Instruction::BinaryOps)Op, RdxParts[part],
+ ReducedPartRdx, "bin.rdx"));
else
ReducedPartRdx = createMinMaxOp(Builder, RdxDesc.MinMaxKind,
ReducedPartRdx, RdxParts[part]);
}
- // VF is a power of 2 so we can emit the reduction using log2(VF) shuffles
- // and vector ops, reducing the set of values being computed by half each
- // round.
- assert(isPowerOf2_32(VF) &&
- "Reduction emission only supported for pow2 vectors!");
- Value *TmpVec = ReducedPartRdx;
- SmallVector<Constant*, 32> ShuffleMask(VF, 0);
- for (unsigned i = VF; i != 1; i >>= 1) {
- // Move the upper half of the vector to the lower half.
- for (unsigned j = 0; j != i/2; ++j)
- ShuffleMask[j] = Builder.getInt32(i/2 + j);
-
- // Fill the rest of the mask with undef.
- std::fill(&ShuffleMask[i/2], ShuffleMask.end(),
- UndefValue::get(Builder.getInt32Ty()));
-
- Value *Shuf =
+ if (VF > 1) {
+ // VF is a power of 2 so we can emit the reduction using log2(VF) shuffles
+ // and vector ops, reducing the set of values being computed by half each
+ // round.
+ assert(isPowerOf2_32(VF) &&
+ "Reduction emission only supported for pow2 vectors!");
+ Value *TmpVec = ReducedPartRdx;
+ SmallVector<Constant*, 32> ShuffleMask(VF, 0);
+ for (unsigned i = VF; i != 1; i >>= 1) {
+ // Move the upper half of the vector to the lower half.
+ for (unsigned j = 0; j != i/2; ++j)
+ ShuffleMask[j] = Builder.getInt32(i/2 + j);
+
+ // Fill the rest of the mask with undef.
+ std::fill(&ShuffleMask[i/2], ShuffleMask.end(),
+ UndefValue::get(Builder.getInt32Ty()));
+
+ Value *Shuf =
Builder.CreateShuffleVector(TmpVec,
UndefValue::get(TmpVec->getType()),
ConstantVector::get(ShuffleMask),
"rdx.shuf");
- if (Op != Instruction::ICmp && Op != Instruction::FCmp)
- TmpVec = Builder.CreateBinOp((Instruction::BinaryOps)Op, TmpVec, Shuf,
- "bin.rdx");
- else
- TmpVec = createMinMaxOp(Builder, RdxDesc.MinMaxKind, TmpVec, Shuf);
- }
+ if (Op != Instruction::ICmp && Op != Instruction::FCmp)
+ // Floating point operations had to be 'fast' to enable the reduction.
+ TmpVec = addFastMathFlag(Builder.CreateBinOp(
+ (Instruction::BinaryOps)Op, TmpVec, Shuf, "bin.rdx"));
+ else
+ TmpVec = createMinMaxOp(Builder, RdxDesc.MinMaxKind, TmpVec, Shuf);
+ }
- // The result is in the first element of the vector.
- Value *Scalar0 = Builder.CreateExtractElement(TmpVec, Builder.getInt32(0));
+ // The result is in the first element of the vector.
+ ReducedPartRdx = Builder.CreateExtractElement(TmpVec,
+ Builder.getInt32(0));
+ }
// Now, we need to fix the users of the reduction variable
// inside and outside of the scalar remainder loop.
for (BasicBlock::iterator LEI = LoopExitBlock->begin(),
LEE = LoopExitBlock->end(); LEI != LEE; ++LEI) {
PHINode *LCSSAPhi = dyn_cast<PHINode>(LEI);
- if (!LCSSAPhi) continue;
+ if (!LCSSAPhi) break;
// All PHINodes need to have a single entry edge, or two if
// we already fixed them.
// incoming bypass edge.
if (LCSSAPhi->getIncomingValue(0) == RdxDesc.LoopExitInstr) {
// Add an edge coming from the bypass.
- LCSSAPhi->addIncoming(Scalar0, LoopMiddleBlock);
+ LCSSAPhi->addIncoming(ReducedPartRdx, LoopMiddleBlock);
break;
}
}// end of the LCSSA phi scan.
assert(IncomingEdgeBlockIdx >= 0 && "Invalid block index");
// Pick the other block.
int SelfEdgeBlockIdx = (IncomingEdgeBlockIdx ? 0 : 1);
- (RdxPhi)->setIncomingValue(SelfEdgeBlockIdx, Scalar0);
+ (RdxPhi)->setIncomingValue(SelfEdgeBlockIdx, ReducedPartRdx);
(RdxPhi)->setIncomingValue(IncomingEdgeBlockIdx, RdxDesc.LoopExitInstr);
}// end of for each redux variable.
- // The Loop exit block may have single value PHI nodes where the incoming
- // value is 'undef'. While vectorizing we only handled real values that
- // were defined inside the loop. Here we handle the 'undef case'.
- // See PR14725.
+ fixLCSSAPHIs();
+
+ // Remove redundant induction instructions.
+ cse(LoopVectorBody);
+}
+
+void InnerLoopVectorizer::fixLCSSAPHIs() {
for (BasicBlock::iterator LEI = LoopExitBlock->begin(),
LEE = LoopExitBlock->end(); LEI != LEE; ++LEI) {
PHINode *LCSSAPhi = dyn_cast<PHINode>(LEI);
- if (!LCSSAPhi) continue;
+ if (!LCSSAPhi) break;
if (LCSSAPhi->getNumIncomingValues() == 1)
LCSSAPhi->addIncoming(UndefValue::get(LCSSAPhi->getType()),
LoopMiddleBlock);
}
-}
+}
InnerLoopVectorizer::VectorParts
InnerLoopVectorizer::createEdgeMask(BasicBlock *Src, BasicBlock *Dst) {
return BlockMask;
}
-void
-InnerLoopVectorizer::vectorizeBlockInLoop(LoopVectorizationLegality *Legal,
- BasicBlock *BB, PhiVector *PV) {
- // For each instruction in the old loop.
- for (BasicBlock::iterator it = BB->begin(), e = BB->end(); it != e; ++it) {
- VectorParts &Entry = WidenMap.get(it);
- switch (it->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>(it);
- // Handle reduction variables:
- if (Legal->getReductionVars()->count(P)) {
- for (unsigned part = 0; part < UF; ++part) {
- // This is phase one of vectorizing PHIs.
- Type *VecTy = VectorType::get(it->getType(), VF);
- Entry[part] = PHINode::Create(VecTy, 2, "vec.phi",
- LoopVectorBody-> getFirstInsertionPt());
- }
- PV->push_back(P);
- continue;
- }
+void InnerLoopVectorizer::widenPHIInstruction(Instruction *PN,
+ InnerLoopVectorizer::VectorParts &Entry,
+ unsigned UF, unsigned VF, PhiVector *PV) {
+ PHINode* P = cast<PHINode>(PN);
+ // Handle reduction variables:
+ if (Legal->getReductionVars()->count(P)) {
+ for (unsigned part = 0; part < UF; ++part) {
+ // This is phase one of vectorizing PHIs.
+ Type *VecTy = (VF == 1) ? PN->getType() :
+ VectorType::get(PN->getType(), VF);
+ Entry[part] = PHINode::Create(VecTy, 2, "vec.phi",
+ LoopVectorBody.back()-> getFirstInsertionPt());
+ }
+ PV->push_back(P);
+ return;
+ }
- setDebugLocFromInst(Builder, P);
- // Check for PHI nodes that are lowered to vector selects.
- if (P->getParent() != OrigLoop->getHeader()) {
- // We know that all PHIs in non header blocks are converted into
- // selects, so we don't have to worry about the insertion order and we
- // can just use the builder.
- // At this point we generate the predication tree. There may be
- // duplications since this is a simple recursive scan, but future
- // optimizations will clean it up.
-
- unsigned NumIncoming = P->getNumIncomingValues();
-
- // Generate a sequence of selects of the form:
- // SELECT(Mask3, In3,
- // SELECT(Mask2, In2,
- // ( ...)))
- for (unsigned In = 0; In < NumIncoming; In++) {
- VectorParts Cond = createEdgeMask(P->getIncomingBlock(In),
- P->getParent());
- VectorParts &In0 = getVectorValue(P->getIncomingValue(In));
-
- for (unsigned part = 0; part < UF; ++part) {
- // We might have single edge PHIs (blocks) - use an identity
- // 'select' for the first PHI operand.
- if (In == 0)
- Entry[part] = Builder.CreateSelect(Cond[part], In0[part],
- In0[part]);
- else
- // Select between the current value and the previous incoming edge
- // based on the incoming mask.
- Entry[part] = Builder.CreateSelect(Cond[part], In0[part],
- Entry[part], "predphi");
- }
- }
- continue;
+ setDebugLocFromInst(Builder, P);
+ // Check for PHI nodes that are lowered to vector selects.
+ if (P->getParent() != OrigLoop->getHeader()) {
+ // We know that all PHIs in non-header blocks are converted into
+ // selects, so we don't have to worry about the insertion order and we
+ // can just use the builder.
+ // At this point we generate the predication tree. There may be
+ // duplications since this is a simple recursive scan, but future
+ // optimizations will clean it up.
+
+ unsigned NumIncoming = P->getNumIncomingValues();
+
+ // Generate a sequence of selects of the form:
+ // SELECT(Mask3, In3,
+ // SELECT(Mask2, In2,
+ // ( ...)))
+ for (unsigned In = 0; In < NumIncoming; In++) {
+ VectorParts Cond = createEdgeMask(P->getIncomingBlock(In),
+ P->getParent());
+ VectorParts &In0 = getVectorValue(P->getIncomingValue(In));
+
+ for (unsigned part = 0; part < UF; ++part) {
+ // We might have single edge PHIs (blocks) - use an identity
+ // 'select' for the first PHI operand.
+ if (In == 0)
+ Entry[part] = Builder.CreateSelect(Cond[part], In0[part],
+ In0[part]);
+ else
+ // Select between the current value and the previous incoming edge
+ // based on the incoming mask.
+ Entry[part] = Builder.CreateSelect(Cond[part], In0[part],
+ Entry[part], "predphi");
}
+ }
+ return;
+ }
- // This PHINode must be an induction variable.
- // Make sure that we know about it.
- assert(Legal->getInductionVars()->count(P) &&
- "Not an induction variable");
-
- LoopVectorizationLegality::InductionInfo II =
- Legal->getInductionVars()->lookup(P);
-
- switch (II.IK) {
- case LoopVectorizationLegality::IK_NoInduction:
- llvm_unreachable("Unknown induction");
- case LoopVectorizationLegality::IK_IntInduction: {
- assert(P->getType() == II.StartValue->getType() && "Types must match");
- Type *PhiTy = P->getType();
- Value *Broadcasted;
- if (P == OldInduction) {
- // Handle the canonical induction variable. We might have had to
- // extend the type.
- Broadcasted = Builder.CreateTrunc(Induction, PhiTy);
- } else {
- // Handle other induction variables that are now based on the
- // canonical one.
- Value *NormalizedIdx = Builder.CreateSub(Induction, ExtendedIdx,
- "normalized.idx");
- NormalizedIdx = Builder.CreateSExtOrTrunc(NormalizedIdx, PhiTy);
- Broadcasted = Builder.CreateAdd(II.StartValue, NormalizedIdx,
- "offset.idx");
- }
- Broadcasted = getBroadcastInstrs(Broadcasted);
- // After broadcasting the induction variable we need to make the vector
- // consecutive by adding 0, 1, 2, etc.
+ // This PHINode must be an induction variable.
+ // Make sure that we know about it.
+ assert(Legal->getInductionVars()->count(P) &&
+ "Not an induction variable");
+
+ LoopVectorizationLegality::InductionInfo II =
+ Legal->getInductionVars()->lookup(P);
+
+ switch (II.IK) {
+ case LoopVectorizationLegality::IK_NoInduction:
+ llvm_unreachable("Unknown induction");
+ case LoopVectorizationLegality::IK_IntInduction: {
+ assert(P->getType() == II.StartValue->getType() && "Types must match");
+ Type *PhiTy = P->getType();
+ Value *Broadcasted;
+ if (P == OldInduction) {
+ // Handle the canonical induction variable. We might have had to
+ // extend the type.
+ Broadcasted = Builder.CreateTrunc(Induction, PhiTy);
+ } else {
+ // Handle other induction variables that are now based on the
+ // canonical one.
+ Value *NormalizedIdx = Builder.CreateSub(Induction, ExtendedIdx,
+ "normalized.idx");
+ NormalizedIdx = Builder.CreateSExtOrTrunc(NormalizedIdx, PhiTy);
+ Broadcasted = Builder.CreateAdd(II.StartValue, NormalizedIdx,
+ "offset.idx");
+ }
+ Broadcasted = getBroadcastInstrs(Broadcasted);
+ // After broadcasting the induction variable we need to make the vector
+ // consecutive by adding 0, 1, 2, etc.
+ for (unsigned part = 0; part < UF; ++part)
+ Entry[part] = getConsecutiveVector(Broadcasted, VF * part, false);
+ return;
+ }
+ case LoopVectorizationLegality::IK_ReverseIntInduction:
+ case LoopVectorizationLegality::IK_PtrInduction:
+ case LoopVectorizationLegality::IK_ReversePtrInduction:
+ // Handle reverse integer and pointer inductions.
+ Value *StartIdx = ExtendedIdx;
+ // This is the normalized GEP that starts counting at zero.
+ Value *NormalizedIdx = Builder.CreateSub(Induction, StartIdx,
+ "normalized.idx");
+
+ // Handle the reverse integer induction variable case.
+ if (LoopVectorizationLegality::IK_ReverseIntInduction == II.IK) {
+ IntegerType *DstTy = cast<IntegerType>(II.StartValue->getType());
+ Value *CNI = Builder.CreateSExtOrTrunc(NormalizedIdx, DstTy,
+ "resize.norm.idx");
+ Value *ReverseInd = Builder.CreateSub(II.StartValue, CNI,
+ "reverse.idx");
+
+ // This is a new value so do not hoist it out.
+ Value *Broadcasted = getBroadcastInstrs(ReverseInd);
+ // After broadcasting the induction variable we need to make the
+ // vector consecutive by adding ... -3, -2, -1, 0.
for (unsigned part = 0; part < UF; ++part)
- Entry[part] = getConsecutiveVector(Broadcasted, VF * part, false);
- continue;
+ Entry[part] = getConsecutiveVector(Broadcasted, -(int)VF * part,
+ true);
+ return;
}
- case LoopVectorizationLegality::IK_ReverseIntInduction:
- case LoopVectorizationLegality::IK_PtrInduction:
- case LoopVectorizationLegality::IK_ReversePtrInduction:
- // Handle reverse integer and pointer inductions.
- Value *StartIdx = ExtendedIdx;
- // This is the normalized GEP that starts counting at zero.
- Value *NormalizedIdx = Builder.CreateSub(Induction, StartIdx,
- "normalized.idx");
- // Handle the reverse integer induction variable case.
- if (LoopVectorizationLegality::IK_ReverseIntInduction == II.IK) {
- IntegerType *DstTy = cast<IntegerType>(II.StartValue->getType());
- Value *CNI = Builder.CreateSExtOrTrunc(NormalizedIdx, DstTy,
- "resize.norm.idx");
- Value *ReverseInd = Builder.CreateSub(II.StartValue, CNI,
- "reverse.idx");
-
- // This is a new value so do not hoist it out.
- Value *Broadcasted = getBroadcastInstrs(ReverseInd);
- // After broadcasting the induction variable we need to make the
- // vector consecutive by adding ... -3, -2, -1, 0.
- for (unsigned part = 0; part < UF; ++part)
- Entry[part] = getConsecutiveVector(Broadcasted, -(int)VF * part,
- true);
+ // Handle the pointer induction variable case.
+ assert(P->getType()->isPointerTy() && "Unexpected type.");
+
+ // Is this a reverse induction ptr or a consecutive induction ptr.
+ bool Reverse = (LoopVectorizationLegality::IK_ReversePtrInduction ==
+ II.IK);
+
+ // This is the vector of results. Notice that we don't generate
+ // vector geps because scalar geps result in better code.
+ for (unsigned part = 0; part < UF; ++part) {
+ if (VF == 1) {
+ int EltIndex = (part) * (Reverse ? -1 : 1);
+ Constant *Idx = ConstantInt::get(Induction->getType(), EltIndex);
+ Value *GlobalIdx;
+ if (Reverse)
+ GlobalIdx = Builder.CreateSub(Idx, NormalizedIdx, "gep.ridx");
+ else
+ GlobalIdx = Builder.CreateAdd(NormalizedIdx, Idx, "gep.idx");
+
+ Value *SclrGep = Builder.CreateGEP(II.StartValue, GlobalIdx,
+ "next.gep");
+ Entry[part] = SclrGep;
continue;
}
- // Handle the pointer induction variable case.
- assert(P->getType()->isPointerTy() && "Unexpected type.");
-
- // Is this a reverse induction ptr or a consecutive induction ptr.
- bool Reverse = (LoopVectorizationLegality::IK_ReversePtrInduction ==
- II.IK);
-
- // This is the vector of results. Notice that we don't generate
- // vector geps because scalar geps result in better code.
- for (unsigned part = 0; part < UF; ++part) {
- Value *VecVal = UndefValue::get(VectorType::get(P->getType(), VF));
- for (unsigned int i = 0; i < VF; ++i) {
- int EltIndex = (i + part * VF) * (Reverse ? -1 : 1);
- Constant *Idx = ConstantInt::get(Induction->getType(), EltIndex);
- Value *GlobalIdx;
- if (!Reverse)
- GlobalIdx = Builder.CreateAdd(NormalizedIdx, Idx, "gep.idx");
- else
- GlobalIdx = Builder.CreateSub(Idx, NormalizedIdx, "gep.ridx");
-
- Value *SclrGep = Builder.CreateGEP(II.StartValue, GlobalIdx,
- "next.gep");
- VecVal = Builder.CreateInsertElement(VecVal, SclrGep,
- Builder.getInt32(i),
- "insert.gep");
- }
- Entry[part] = VecVal;
+ Value *VecVal = UndefValue::get(VectorType::get(P->getType(), VF));
+ for (unsigned int i = 0; i < VF; ++i) {
+ int EltIndex = (i + part * VF) * (Reverse ? -1 : 1);
+ Constant *Idx = ConstantInt::get(Induction->getType(), EltIndex);
+ Value *GlobalIdx;
+ if (!Reverse)
+ GlobalIdx = Builder.CreateAdd(NormalizedIdx, Idx, "gep.idx");
+ else
+ GlobalIdx = Builder.CreateSub(Idx, NormalizedIdx, "gep.ridx");
+
+ Value *SclrGep = Builder.CreateGEP(II.StartValue, GlobalIdx,
+ "next.gep");
+ VecVal = Builder.CreateInsertElement(VecVal, SclrGep,
+ Builder.getInt32(i),
+ "insert.gep");
}
- continue;
+ Entry[part] = VecVal;
}
+ return;
+ }
+}
+void InnerLoopVectorizer::vectorizeBlockInLoop(BasicBlock *BB, PhiVector *PV) {
+ // For each instruction in the old loop.
+ for (BasicBlock::iterator it = BB->begin(), e = BB->end(); it != e; ++it) {
+ VectorParts &Entry = WidenMap.get(it);
+ switch (it->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:{
+ // Vectorize PHINodes.
+ widenPHIInstruction(it, Entry, UF, VF, PV);
+ continue;
}// End of PHI.
case Instruction::Add:
if (VecOp && isa<PossiblyExactOperator>(VecOp))
VecOp->setIsExact(BinOp->isExact());
+ // Copy the fast-math flags.
+ if (VecOp && isa<FPMathOperator>(V))
+ VecOp->setFastMathFlags(it->getFastMathFlags());
+
Entry[Part] = V;
}
break;
VectorParts &Cond = getVectorValue(it->getOperand(0));
VectorParts &Op0 = getVectorValue(it->getOperand(1));
VectorParts &Op1 = getVectorValue(it->getOperand(2));
- Value *ScalarCond = Builder.CreateExtractElement(Cond[0],
- Builder.getInt32(0));
+
+ Value *ScalarCond = (VF == 1) ? Cond[0] :
+ Builder.CreateExtractElement(Cond[0], Builder.getInt32(0));
+
for (unsigned Part = 0; Part < UF; ++Part) {
Entry[Part] = Builder.CreateSelect(
InvariantCond ? ScalarCond : Cond[Part],
case Instruction::Store:
case Instruction::Load:
- vectorizeMemoryInstruction(it, Legal);
+ vectorizeMemoryInstruction(it);
break;
case Instruction::ZExt:
case Instruction::SExt:
break;
}
/// Vectorize casts.
- Type *DestTy = VectorType::get(CI->getType()->getScalarType(), VF);
+ Type *DestTy = (VF == 1) ? CI->getType() :
+ VectorType::get(CI->getType(), VF);
VectorParts &A = getVectorValue(it->getOperand(0));
for (unsigned Part = 0; Part < UF; ++Part)
CallInst *CI = cast<CallInst>(it);
Intrinsic::ID ID = getIntrinsicIDForCall(CI, TLI);
assert(ID && "Not an intrinsic call!");
- for (unsigned Part = 0; Part < UF; ++Part) {
- SmallVector<Value*, 4> Args;
- for (unsigned i = 0, ie = CI->getNumArgOperands(); i != ie; ++i) {
- VectorParts &Arg = getVectorValue(CI->getArgOperand(i));
- Args.push_back(Arg[Part]);
+ switch (ID) {
+ case Intrinsic::lifetime_end:
+ case Intrinsic::lifetime_start:
+ scalarizeInstruction(it);
+ break;
+ default:
+ for (unsigned Part = 0; Part < UF; ++Part) {
+ SmallVector<Value *, 4> Args;
+ for (unsigned i = 0, ie = CI->getNumArgOperands(); i != ie; ++i) {
+ VectorParts &Arg = getVectorValue(CI->getArgOperand(i));
+ Args.push_back(Arg[Part]);
+ }
+ Type *Tys[] = {CI->getType()};
+ if (VF > 1)
+ Tys[0] = VectorType::get(CI->getType()->getScalarType(), VF);
+
+ Function *F = Intrinsic::getDeclaration(M, ID, Tys);
+ Entry[Part] = Builder.CreateCall(F, Args);
}
- Type *Tys[] = { VectorType::get(CI->getType()->getScalarType(), VF) };
- Function *F = Intrinsic::getDeclaration(M, ID, Tys);
- Entry[Part] = Builder.CreateCall(F, Args);
+ break;
}
break;
}
for (unsigned I = 1, E = LoopBypassBlocks.size(); I != E; ++I)
DT->addNewBlock(LoopBypassBlocks[I], LoopBypassBlocks[I-1]);
DT->addNewBlock(LoopVectorPreHeader, LoopBypassBlocks.back());
- DT->addNewBlock(LoopVectorBody, LoopVectorPreHeader);
+
+ // Due to if predication of stores we might create a sequence of "if(pred)
+ // a[i] = ...; " blocks.
+ for (unsigned i = 0, e = LoopVectorBody.size(); i != e; ++i) {
+ if (i == 0)
+ DT->addNewBlock(LoopVectorBody[0], LoopVectorPreHeader);
+ else if (isPredicatedBlock(i)) {
+ DT->addNewBlock(LoopVectorBody[i], LoopVectorBody[i-1]);
+ } else {
+ DT->addNewBlock(LoopVectorBody[i], LoopVectorBody[i-2]);
+ }
+ }
+
DT->addNewBlock(LoopMiddleBlock, LoopBypassBlocks.front());
DT->addNewBlock(LoopScalarPreHeader, LoopMiddleBlock);
DT->changeImmediateDominator(LoopScalarBody, LoopScalarPreHeader);
DT->changeImmediateDominator(LoopExitBlock, LoopMiddleBlock);
- DEBUG(DT->verifyAnalysis());
+ DEBUG(DT->verifyDomTree());
+}
+
+/// \brief Check whether it is safe to if-convert this phi node.
+///
+/// Phi nodes with constant expressions that can trap are not safe to if
+/// convert.
+static bool canIfConvertPHINodes(BasicBlock *BB) {
+ for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E; ++I) {
+ PHINode *Phi = dyn_cast<PHINode>(I);
+ if (!Phi)
+ return true;
+ for (unsigned p = 0, e = Phi->getNumIncomingValues(); p != e; ++p)
+ if (Constant *C = dyn_cast<Constant>(Phi->getIncomingValue(p)))
+ if (C->canTrap())
+ return false;
+ }
+ return true;
}
bool LoopVectorizationLegality::canVectorizeWithIfConvert() {
return false;
assert(TheLoop->getNumBlocks() > 1 && "Single block loops are vectorizable");
- std::vector<BasicBlock*> &LoopBlocks = TheLoop->getBlocksVector();
// A list of pointers that we can safely read and write to.
SmallPtrSet<Value *, 8> SafePointes;
// Collect safe addresses.
- for (unsigned i = 0, e = LoopBlocks.size(); i < e; ++i) {
- BasicBlock *BB = LoopBlocks[i];
+ for (Loop::block_iterator BI = TheLoop->block_begin(),
+ BE = TheLoop->block_end(); BI != BE; ++BI) {
+ BasicBlock *BB = *BI;
if (blockNeedsPredication(BB))
continue;
}
// Collect the blocks that need predication.
- for (unsigned i = 0, e = LoopBlocks.size(); i < e; ++i) {
- BasicBlock *BB = LoopBlocks[i];
+ BasicBlock *Header = TheLoop->getHeader();
+ for (Loop::block_iterator BI = TheLoop->block_begin(),
+ BE = TheLoop->block_end(); BI != BE; ++BI) {
+ BasicBlock *BB = *BI;
// We don't support switch statements inside loops.
if (!isa<BranchInst>(BB->getTerminator()))
return false;
// We must be able to predicate all blocks that need to be predicated.
- if (blockNeedsPredication(BB) && !blockCanBePredicated(BB, SafePointes))
+ if (blockNeedsPredication(BB)) {
+ if (!blockCanBePredicated(BB, SafePointes))
+ return false;
+ } else if (BB != Header && !canIfConvertPHINodes(BB))
return false;
+
}
// We can if-convert this loop.
if (!TheLoop->getExitingBlock())
return false;
- unsigned NumBlocks = TheLoop->getNumBlocks();
+ // We need to have a loop header.
+ DEBUG(dbgs() << "LV: Found a loop: " <<
+ TheLoop->getHeader()->getName() << '\n');
- // Check if we can if-convert non single-bb loops.
+ // Check if we can if-convert non-single-bb loops.
+ unsigned NumBlocks = TheLoop->getNumBlocks();
if (NumBlocks != 1 && !canVectorizeWithIfConvert()) {
DEBUG(dbgs() << "LV: Can't if-convert the loop.\n");
return false;
}
- // We need to have a loop header.
- BasicBlock *Latch = TheLoop->getLoopLatch();
- DEBUG(dbgs() << "LV: Found a loop: " <<
- TheLoop->getHeader()->getName() << "\n");
-
// ScalarEvolution needs to be able to find the exit count.
const SCEV *ExitCount = SE->getBackedgeTakenCount(TheLoop);
if (ExitCount == SE->getCouldNotCompute()) {
}
// Do not loop-vectorize loops with a tiny trip count.
+ BasicBlock *Latch = TheLoop->getLoopLatch();
unsigned TC = SE->getSmallConstantTripCount(TheLoop, Latch);
if (TC > 0u && TC < TinyTripCountVectorThreshold) {
DEBUG(dbgs() << "LV: Found a loop with a very small trip count. " <<
return true;
}
-static Type *convertPointerToIntegerType(DataLayout &DL, Type *Ty) {
+static Type *convertPointerToIntegerType(const DataLayout &DL, Type *Ty) {
if (Ty->isPointerTy())
- return DL.getIntPtrType(Ty->getContext());
+ return DL.getIntPtrType(Ty);
+
+ // It is possible that char's or short's overflow when we ask for the loop's
+ // trip count, work around this by changing the type size.
+ if (Ty->getScalarSizeInBits() < 32)
+ return Type::getInt32Ty(Ty->getContext());
+
return Ty;
}
-static Type* getWiderType(DataLayout &DL, Type *Ty0, Type *Ty1) {
+static Type* getWiderType(const DataLayout &DL, Type *Ty0, Type *Ty1) {
Ty0 = convertPointerToIntegerType(DL, Ty0);
Ty1 = convertPointerToIntegerType(DL, Ty1);
if (Ty0->getScalarSizeInBits() > Ty1->getScalarSizeInBits())
// instructions must not have external users.
if (!Reductions.count(Inst))
//Check that all of the users of the loop are inside the BB.
- for (Value::use_iterator I = Inst->use_begin(), E = Inst->use_end();
- I != E; ++I) {
- Instruction *U = cast<Instruction>(*I);
+ for (User *U : Inst->users()) {
+ Instruction *UI = cast<Instruction>(U);
// This user may be a reduction exit value.
- if (!TheLoop->contains(U)) {
- DEBUG(dbgs() << "LV: Found an outside user for : "<< *U << "\n");
+ if (!TheLoop->contains(UI)) {
+ DEBUG(dbgs() << "LV: Found an outside user for : " << *UI << '\n');
return true;
}
}
DEBUG(dbgs() << "LV: Found an induction variable.\n");
Inductions[Phi] = InductionInfo(StartValue, IK);
+
+ // Until we explicitly handle the case of an induction variable with
+ // an outside loop user we have to give up vectorizing this loop.
+ if (hasOutsideLoopUser(TheLoop, it, AllowedExit))
+ return false;
+
continue;
}
continue;
}
if (AddReductionVar(Phi, RK_FloatMinMax)) {
- DEBUG(dbgs() << "LV: Found an float MINMAX reduction PHI."<< *Phi <<"\n");
+ DEBUG(dbgs() << "LV: Found an float MINMAX reduction PHI."<< *Phi <<
+ "\n");
continue;
}
}
// Check that the instruction return type is vectorizable.
- if (!VectorType::isValidElementType(it->getType()) &&
- !it->getType()->isVoidTy()) {
- DEBUG(dbgs() << "LV: Found unvectorizable type." << "\n");
+ // Also, we can't vectorize extractelement instructions.
+ if ((!VectorType::isValidElementType(it->getType()) &&
+ !it->getType()->isVoidTy()) || isa<ExtractElementInst>(it)) {
+ DEBUG(dbgs() << "LV: Found unvectorizable type.\n");
return false;
}
Type *T = ST->getValueOperand()->getType();
if (!VectorType::isValidElementType(T))
return false;
+ if (EnableMemAccessVersioning)
+ collectStridedAcccess(ST);
}
+ if (EnableMemAccessVersioning)
+ if (LoadInst *LI = dyn_cast<LoadInst>(it))
+ collectStridedAcccess(LI);
+
// Reduction instructions are allowed to have exit users.
// All other instructions must not have external users.
if (hasOutsideLoopUser(TheLoop, it, AllowedExit))
return true;
}
+///\brief Remove GEPs whose indices but the last one are loop invariant and
+/// return the induction operand of the gep pointer.
+static Value *stripGetElementPtr(Value *Ptr, ScalarEvolution *SE,
+ const DataLayout *DL, Loop *Lp) {
+ GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(Ptr);
+ if (!GEP)
+ return Ptr;
+
+ unsigned InductionOperand = getGEPInductionOperand(DL, GEP);
+
+ // Check that all of the gep indices are uniform except for our induction
+ // operand.
+ for (unsigned i = 0, e = GEP->getNumOperands(); i != e; ++i)
+ if (i != InductionOperand &&
+ !SE->isLoopInvariant(SE->getSCEV(GEP->getOperand(i)), Lp))
+ return Ptr;
+ return GEP->getOperand(InductionOperand);
+}
+
+///\brief Look for a cast use of the passed value.
+static Value *getUniqueCastUse(Value *Ptr, Loop *Lp, Type *Ty) {
+ Value *UniqueCast = 0;
+ for (User *U : Ptr->users()) {
+ CastInst *CI = dyn_cast<CastInst>(U);
+ if (CI && CI->getType() == Ty) {
+ if (!UniqueCast)
+ UniqueCast = CI;
+ else
+ return 0;
+ }
+ }
+ return UniqueCast;
+}
+
+///\brief Get the stride of a pointer access in a loop.
+/// Looks for symbolic strides "a[i*stride]". Returns the symbolic stride as a
+/// pointer to the Value, or null otherwise.
+static Value *getStrideFromPointer(Value *Ptr, ScalarEvolution *SE,
+ const DataLayout *DL, Loop *Lp) {
+ const PointerType *PtrTy = dyn_cast<PointerType>(Ptr->getType());
+ if (!PtrTy || PtrTy->isAggregateType())
+ return 0;
+
+ // Try to remove a gep instruction to make the pointer (actually index at this
+ // point) easier analyzable. If OrigPtr is equal to Ptr we are analzying the
+ // pointer, otherwise, we are analyzing the index.
+ Value *OrigPtr = Ptr;
+
+ // The size of the pointer access.
+ int64_t PtrAccessSize = 1;
+
+ Ptr = stripGetElementPtr(Ptr, SE, DL, Lp);
+ const SCEV *V = SE->getSCEV(Ptr);
+
+ if (Ptr != OrigPtr)
+ // Strip off casts.
+ while (const SCEVCastExpr *C = dyn_cast<SCEVCastExpr>(V))
+ V = C->getOperand();
+
+ const SCEVAddRecExpr *S = dyn_cast<SCEVAddRecExpr>(V);
+ if (!S)
+ return 0;
+
+ V = S->getStepRecurrence(*SE);
+ if (!V)
+ return 0;
+
+ // Strip off the size of access multiplication if we are still analyzing the
+ // pointer.
+ if (OrigPtr == Ptr) {
+ DL->getTypeAllocSize(PtrTy->getElementType());
+ if (const SCEVMulExpr *M = dyn_cast<SCEVMulExpr>(V)) {
+ if (M->getOperand(0)->getSCEVType() != scConstant)
+ return 0;
+
+ const APInt &APStepVal =
+ cast<SCEVConstant>(M->getOperand(0))->getValue()->getValue();
+
+ // Huge step value - give up.
+ if (APStepVal.getBitWidth() > 64)
+ return 0;
+
+ int64_t StepVal = APStepVal.getSExtValue();
+ if (PtrAccessSize != StepVal)
+ return 0;
+ V = M->getOperand(1);
+ }
+ }
+
+ // Strip off casts.
+ Type *StripedOffRecurrenceCast = 0;
+ if (const SCEVCastExpr *C = dyn_cast<SCEVCastExpr>(V)) {
+ StripedOffRecurrenceCast = C->getType();
+ V = C->getOperand();
+ }
+
+ // Look for the loop invariant symbolic value.
+ const SCEVUnknown *U = dyn_cast<SCEVUnknown>(V);
+ if (!U)
+ return 0;
+
+ Value *Stride = U->getValue();
+ if (!Lp->isLoopInvariant(Stride))
+ return 0;
+
+ // If we have stripped off the recurrence cast we have to make sure that we
+ // return the value that is used in this loop so that we can replace it later.
+ if (StripedOffRecurrenceCast)
+ Stride = getUniqueCastUse(Stride, Lp, StripedOffRecurrenceCast);
+
+ return Stride;
+}
+
+void LoopVectorizationLegality::collectStridedAcccess(Value *MemAccess) {
+ Value *Ptr = 0;
+ if (LoadInst *LI = dyn_cast<LoadInst>(MemAccess))
+ Ptr = LI->getPointerOperand();
+ else if (StoreInst *SI = dyn_cast<StoreInst>(MemAccess))
+ Ptr = SI->getPointerOperand();
+ else
+ return;
+
+ Value *Stride = getStrideFromPointer(Ptr, SE, DL, TheLoop);
+ if (!Stride)
+ return;
+
+ DEBUG(dbgs() << "LV: Found a strided access that we can version");
+ DEBUG(dbgs() << " Ptr: " << *Ptr << " Stride: " << *Stride << "\n");
+ Strides[Ptr] = Stride;
+ StrideSet.insert(Stride);
+}
+
void LoopVectorizationLegality::collectLoopUniforms() {
// We now know that the loop is vectorizable!
// Collect variables that will remain uniform after vectorization.
// Start with the conditional branch and walk up the block.
Worklist.push_back(Latch->getTerminator()->getOperand(0));
+ // Also add all consecutive pointer values; these values will be uniform
+ // after vectorization (and subsequent cleanup) and, until revectorization is
+ // supported, all dependencies must also be uniform.
+ for (Loop::block_iterator B = TheLoop->block_begin(),
+ BE = TheLoop->block_end(); B != BE; ++B)
+ for (BasicBlock::iterator I = (*B)->begin(), IE = (*B)->end();
+ I != IE; ++I)
+ if (I->getType()->isPointerTy() && isConsecutivePtr(I))
+ Worklist.insert(Worklist.end(), I->op_begin(), I->op_end());
+
while (Worklist.size()) {
Instruction *I = dyn_cast<Instruction>(Worklist.back());
Worklist.pop_back();
/// \brief Set of potential dependent memory accesses.
typedef EquivalenceClasses<MemAccessInfo> DepCandidates;
- AccessAnalysis(DataLayout *Dl, DepCandidates &DA) :
+ AccessAnalysis(const DataLayout *Dl, DepCandidates &DA) :
DL(Dl), DepCands(DA), AreAllWritesIdentified(true),
AreAllReadsIdentified(true), IsRTCheckNeeded(false) {}
/// non-intersection.
bool canCheckPtrAtRT(LoopVectorizationLegality::RuntimePointerCheck &RtCheck,
unsigned &NumComparisons, ScalarEvolution *SE,
- Loop *TheLoop);
+ Loop *TheLoop, ValueToValueMap &Strides,
+ bool ShouldCheckStride = false);
/// \brief Goes over all memory accesses, checks whether a RT check is needed
/// and builds sets of dependent accesses.
bool isRTCheckNeeded() { return IsRTCheckNeeded; }
bool isDependencyCheckNeeded() { return !CheckDeps.empty(); }
+ void resetDepChecks() { CheckDeps.clear(); }
MemAccessInfoSet &getDependenciesToCheck() { return CheckDeps; }
/// Set of underlying objects already written to.
SmallPtrSet<Value*, 16> WriteObjects;
- DataLayout *DL;
+ const DataLayout *DL;
/// Sets of potentially dependent accesses - members of one set share an
/// underlying pointer. The set "CheckDeps" identfies which sets really need a
} // end anonymous namespace
/// \brief Check whether a pointer can participate in a runtime bounds check.
-static bool hasComputableBounds(ScalarEvolution *SE, Value *Ptr) {
- const SCEV *PtrScev = SE->getSCEV(Ptr);
+static bool hasComputableBounds(ScalarEvolution *SE, ValueToValueMap &Strides,
+ Value *Ptr) {
+ const SCEV *PtrScev = replaceSymbolicStrideSCEV(SE, Strides, Ptr);
const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(PtrScev);
if (!AR)
return false;
return AR->isAffine();
}
+/// \brief Check the stride of the pointer and ensure that it does not wrap in
+/// the address space.
+static int isStridedPtr(ScalarEvolution *SE, const DataLayout *DL, Value *Ptr,
+ const Loop *Lp, ValueToValueMap &StridesMap);
+
bool AccessAnalysis::canCheckPtrAtRT(
- LoopVectorizationLegality::RuntimePointerCheck &RtCheck,
- unsigned &NumComparisons, ScalarEvolution *SE,
- Loop *TheLoop) {
+ LoopVectorizationLegality::RuntimePointerCheck &RtCheck,
+ unsigned &NumComparisons, ScalarEvolution *SE, Loop *TheLoop,
+ ValueToValueMap &StridesMap, bool ShouldCheckStride) {
// Find pointers with computable bounds. We are going to use this information
// to place a runtime bound check.
unsigned NumReadPtrChecks = 0;
else
++NumReadPtrChecks;
- if (hasComputableBounds(SE, Ptr)) {
+ if (hasComputableBounds(SE, StridesMap, Ptr) &&
+ // When we run after a failing dependency check we have to make sure we
+ // don't have wrapping pointers.
+ (!ShouldCheckStride ||
+ isStridedPtr(SE, DL, Ptr, TheLoop, StridesMap) == 1)) {
// The id of the dependence set.
unsigned DepId;
// Each access has its own dependence set.
DepId = RunningDepId++;
- RtCheck.insert(SE, TheLoop, Ptr, IsWrite, DepId);
+ RtCheck.insert(SE, TheLoop, Ptr, IsWrite, DepId, StridesMap);
- DEBUG(dbgs() << "LV: Found a runtime check ptr:" << *Ptr <<"\n");
+ DEBUG(dbgs() << "LV: Found a runtime check ptr:" << *Ptr << '\n');
} else {
CanDoRT = false;
}
if (IsDepCheckNeeded && CanDoRT && RunningDepId == 2)
NumComparisons = 0; // Only one dependence set.
- else
+ else {
NumComparisons = (NumWritePtrChecks * (NumReadPtrChecks +
NumWritePtrChecks - 1));
+ }
+
+ // If the pointers that we would use for the bounds comparison have different
+ // address spaces, assume the values aren't directly comparable, so we can't
+ // use them for the runtime check. We also have to assume they could
+ // overlap. In the future there should be metadata for whether address spaces
+ // are disjoint.
+ unsigned NumPointers = RtCheck.Pointers.size();
+ for (unsigned i = 0; i < NumPointers; ++i) {
+ for (unsigned j = i + 1; j < NumPointers; ++j) {
+ // Only need to check pointers between two different dependency sets.
+ if (RtCheck.DependencySetId[i] == RtCheck.DependencySetId[j])
+ continue;
+
+ Value *PtrI = RtCheck.Pointers[i];
+ Value *PtrJ = RtCheck.Pointers[j];
+
+ unsigned ASi = PtrI->getType()->getPointerAddressSpace();
+ unsigned ASj = PtrJ->getType()->getPointerAddressSpace();
+ if (ASi != ASj) {
+ DEBUG(dbgs() << "LV: Runtime check would require comparison between"
+ " different address spaces\n");
+ return false;
+ }
+ }
+ }
+
return CanDoRT;
}
}
bool NeedDepCheck = false;
- // Check whether there is the possiblity of dependency because of underlying
- // objects being the same.
+ // Check whether there is the possibility of dependency because of
+ // underlying objects being the same.
typedef SmallVector<Value*, 16> ValueVector;
ValueVector TempObjects;
GetUnderlyingObjects(Ptr, TempObjects, DL);
!isa<Argument>(UnderlyingObj)) &&
!isIdentifiedObject(UnderlyingObj))) {
DEBUG(dbgs() << "LV: Found an unidentified " <<
- (IsWrite ? "write" : "read" ) << " ptr:" << *UnderlyingObj <<
+ (IsWrite ? "write" : "read" ) << " ptr: " << *UnderlyingObj <<
"\n");
IsRTCheckNeeded = (IsRTCheckNeeded ||
!isIdentifiedObject(UnderlyingObj) ||
typedef PointerIntPair<Value *, 1, bool> MemAccessInfo;
typedef SmallPtrSet<MemAccessInfo, 8> MemAccessInfoSet;
- MemoryDepChecker(ScalarEvolution *Se, DataLayout *Dl, const Loop *L) :
- SE(Se), DL(Dl), InnermostLoop(L), AccessIdx(0) {}
+ MemoryDepChecker(ScalarEvolution *Se, const DataLayout *Dl, const Loop *L)
+ : SE(Se), DL(Dl), InnermostLoop(L), AccessIdx(0),
+ ShouldRetryWithRuntimeCheck(false) {}
/// \brief Register the location (instructions are given increasing numbers)
/// of a write access.
///
/// Only checks sets with elements in \p CheckDeps.
bool areDepsSafe(AccessAnalysis::DepCandidates &AccessSets,
- MemAccessInfoSet &CheckDeps);
+ MemAccessInfoSet &CheckDeps, ValueToValueMap &Strides);
/// \brief The maximum number of bytes of a vector register we can vectorize
/// the accesses safely with.
unsigned getMaxSafeDepDistBytes() { return MaxSafeDepDistBytes; }
+ /// \brief In same cases when the dependency check fails we can still
+ /// vectorize the loop with a dynamic array access check.
+ bool shouldRetryWithRuntimeCheck() { return ShouldRetryWithRuntimeCheck; }
+
private:
ScalarEvolution *SE;
- DataLayout *DL;
+ const DataLayout *DL;
const Loop *InnermostLoop;
/// \brief Maps access locations (ptr, read/write) to program order.
// We can access this many bytes in parallel safely.
unsigned MaxSafeDepDistBytes;
+ /// \brief If we see a non-constant dependence distance we can still try to
+ /// vectorize this loop with runtime checks.
+ bool ShouldRetryWithRuntimeCheck;
+
/// \brief Check whether there is a plausible dependence between the two
/// accesses.
///
/// distance is smaller than any other distance encountered so far).
/// Otherwise, this function returns true signaling a possible dependence.
bool isDependent(const MemAccessInfo &A, unsigned AIdx,
- const MemAccessInfo &B, unsigned BIdx);
+ const MemAccessInfo &B, unsigned BIdx,
+ ValueToValueMap &Strides);
/// \brief Check whether the data dependence could prevent store-load
/// forwarding.
}
/// \brief Check whether the access through \p Ptr has a constant stride.
-static int isStridedPtr(ScalarEvolution *SE, DataLayout *DL, Value *Ptr,
- const Loop *Lp) {
+static int isStridedPtr(ScalarEvolution *SE, const DataLayout *DL, Value *Ptr,
+ const Loop *Lp, ValueToValueMap &StridesMap) {
const Type *Ty = Ptr->getType();
- assert(Ty->isPointerTy() && "Unexpected non ptr");
+ assert(Ty->isPointerTy() && "Unexpected non-ptr");
// Make sure that the pointer does not point to aggregate types.
const PointerType *PtrTy = cast<PointerType>(Ty);
if (PtrTy->getElementType()->isAggregateType()) {
- DEBUG(dbgs() << "LV: Bad stride - Not a pointer to a scalar type" << *Ptr
- << "\n");
+ DEBUG(dbgs() << "LV: Bad stride - Not a pointer to a scalar type" << *Ptr <<
+ "\n");
return 0;
}
- const SCEV *PtrScev = SE->getSCEV(Ptr);
+ const SCEV *PtrScev = replaceSymbolicStrideSCEV(SE, StridesMap, Ptr);
+
const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(PtrScev);
if (!AR) {
DEBUG(dbgs() << "LV: Bad stride - Not an AddRecExpr pointer "
// The accesss function must stride over the innermost loop.
if (Lp != AR->getLoop()) {
- DEBUG(dbgs() << "LV: Bad stride - Not striding over innermost loop " << *Ptr
- << " SCEV: " << *PtrScev << "\n");
+ DEBUG(dbgs() << "LV: Bad stride - Not striding over innermost loop " <<
+ *Ptr << " SCEV: " << *PtrScev << "\n");
}
// The address calculation must not wrap. Otherwise, a dependence could be
}
bool MemoryDepChecker::isDependent(const MemAccessInfo &A, unsigned AIdx,
- const MemAccessInfo &B, unsigned BIdx) {
+ const MemAccessInfo &B, unsigned BIdx,
+ ValueToValueMap &Strides) {
assert (AIdx < BIdx && "Must pass arguments in program order");
Value *APtr = A.getPointer();
if (!AIsWrite && !BIsWrite)
return false;
- const SCEV *AScev = SE->getSCEV(APtr);
- const SCEV *BScev = SE->getSCEV(BPtr);
+ const SCEV *AScev = replaceSymbolicStrideSCEV(SE, Strides, APtr);
+ const SCEV *BScev = replaceSymbolicStrideSCEV(SE, Strides, BPtr);
- int StrideAPtr = isStridedPtr(SE, DL, APtr, InnermostLoop);
- int StrideBPtr = isStridedPtr(SE, DL, BPtr, InnermostLoop);
+ int StrideAPtr = isStridedPtr(SE, DL, APtr, InnermostLoop, Strides);
+ int StrideBPtr = isStridedPtr(SE, DL, BPtr, InnermostLoop, Strides);
const SCEV *Src = AScev;
const SCEV *Sink = BScev;
const SCEVConstant *C = dyn_cast<SCEVConstant>(Dist);
if (!C) {
- DEBUG(dbgs() << "LV: Dependence because of non constant distance\n");
+ DEBUG(dbgs() << "LV: Dependence because of non-constant distance\n");
+ ShouldRetryWithRuntimeCheck = true;
return true;
}
if (Val == 0) {
if (ATy == BTy)
return false;
- DEBUG(dbgs() << "LV: Zero dependence difference but different types");
+ DEBUG(dbgs() << "LV: Zero dependence difference but different types\n");
return true;
}
// Positive distance bigger than max vectorization factor.
if (ATy != BTy) {
DEBUG(dbgs() <<
- "LV: ReadWrite-Write positive dependency with different types");
+ "LV: ReadWrite-Write positive dependency with different types\n");
return false;
}
2*TypeByteSize > MaxSafeDepDistBytes ||
Distance < TypeByteSize * ForcedUnroll * ForcedFactor) {
DEBUG(dbgs() << "LV: Failure because of Positive distance "
- << Val.getSExtValue() << "\n");
+ << Val.getSExtValue() << '\n');
return true;
}
return true;
DEBUG(dbgs() << "LV: Positive distance " << Val.getSExtValue() <<
- " with max VF=" << MaxSafeDepDistBytes/TypeByteSize << "\n");
+ " with max VF = " << MaxSafeDepDistBytes / TypeByteSize << '\n');
return false;
}
-bool
-MemoryDepChecker::areDepsSafe(AccessAnalysis::DepCandidates &AccessSets,
- MemAccessInfoSet &CheckDeps) {
+bool MemoryDepChecker::areDepsSafe(AccessAnalysis::DepCandidates &AccessSets,
+ MemAccessInfoSet &CheckDeps,
+ ValueToValueMap &Strides) {
MaxSafeDepDistBytes = -1U;
while (!CheckDeps.empty()) {
// Check every access pair.
while (AI != AE) {
CheckDeps.erase(*AI);
- EquivalenceClasses<MemAccessInfo>::member_iterator OI = llvm::next(AI);
+ EquivalenceClasses<MemAccessInfo>::member_iterator OI = std::next(AI);
while (OI != AE) {
// Check every accessing instruction pair in program order.
for (std::vector<unsigned>::iterator I1 = Accesses[*AI].begin(),
I1E = Accesses[*AI].end(); I1 != I1E; ++I1)
for (std::vector<unsigned>::iterator I2 = Accesses[*OI].begin(),
I2E = Accesses[*OI].end(); I2 != I2E; ++I2) {
- if (*I1 < *I2 && isDependent(*AI, *I1, *OI, *I2))
+ if (*I1 < *I2 && isDependent(*AI, *I1, *OI, *I2, Strides))
return false;
- if (*I2 < *I1 && isDependent(*OI, *I2, *AI, *I1))
+ if (*I2 < *I1 && isDependent(*OI, *I2, *AI, *I1, Strides))
return false;
}
++OI;
DEBUG(dbgs() << "LV: Found a non-simple load.\n");
return false;
}
+ NumLoads++;
Loads.push_back(Ld);
DepChecker.addAccess(Ld);
continue;
DEBUG(dbgs() << "LV: Found a non-simple store.\n");
return false;
}
+ NumStores++;
Stores.push_back(St);
DepChecker.addAccess(St);
}
- } // next instr.
- } // next block.
+ } // Next instr.
+ } // Next block.
// Now we have two lists that hold the loads and the stores.
// Next, we find the pointers that they use.
return true;
}
- SmallPtrSet<Value *, 16> ReadOnlyPtr;
for (I = Loads.begin(), IE = Loads.end(); I != IE; ++I) {
LoadInst *LD = cast<LoadInst>(*I);
Value* Ptr = LD->getPointerOperand();
// read a few words, modify, and write a few words, and some of the
// words may be written to the same address.
bool IsReadOnlyPtr = false;
- if (Seen.insert(Ptr) || !isStridedPtr(SE, DL, Ptr, TheLoop)) {
+ if (Seen.insert(Ptr) || !isStridedPtr(SE, DL, Ptr, TheLoop, Strides)) {
++NumReads;
IsReadOnlyPtr = true;
}
unsigned NumComparisons = 0;
bool CanDoRT = false;
if (NeedRTCheck)
- CanDoRT = Accesses.canCheckPtrAtRT(PtrRtCheck, NumComparisons, SE, TheLoop);
-
+ CanDoRT = Accesses.canCheckPtrAtRT(PtrRtCheck, NumComparisons, SE, TheLoop,
+ Strides);
DEBUG(dbgs() << "LV: We need to do " << NumComparisons <<
" pointer comparisons.\n");
if (NumComparisons == 0 && NeedRTCheck)
NeedRTCheck = false;
- // Check that we did not collect too many pointers or found a unsizeable
+ // Check that we did not collect too many pointers or found an unsizeable
// pointer.
if (!CanDoRT || NumComparisons > RuntimeMemoryCheckThreshold) {
PtrRtCheck.reset();
bool CanVecMem = true;
if (Accesses.isDependencyCheckNeeded()) {
DEBUG(dbgs() << "LV: Checking memory dependencies\n");
- CanVecMem = DepChecker.areDepsSafe(DependentAccesses,
- Accesses.getDependenciesToCheck());
+ CanVecMem = DepChecker.areDepsSafe(
+ DependentAccesses, Accesses.getDependenciesToCheck(), Strides);
MaxSafeDepDistBytes = DepChecker.getMaxSafeDepDistBytes();
+
+ if (!CanVecMem && DepChecker.shouldRetryWithRuntimeCheck()) {
+ DEBUG(dbgs() << "LV: Retrying with memory checks\n");
+ NeedRTCheck = true;
+
+ // Clear the dependency checks. We assume they are not needed.
+ Accesses.resetDepChecks();
+
+ PtrRtCheck.reset();
+ PtrRtCheck.Need = true;
+
+ CanDoRT = Accesses.canCheckPtrAtRT(PtrRtCheck, NumComparisons, SE,
+ TheLoop, Strides, true);
+ // Check that we did not collect too many pointers or found an unsizeable
+ // pointer.
+ if (!CanDoRT || NumComparisons > RuntimeMemoryCheckThreshold) {
+ DEBUG(dbgs() << "LV: Can't vectorize with memory checks\n");
+ PtrRtCheck.reset();
+ return false;
+ }
+
+ CanVecMem = true;
+ }
}
- DEBUG(dbgs() << "LV: We "<< (NeedRTCheck ? "" : "don't") <<
+ DEBUG(dbgs() << "LV: We" << (NeedRTCheck ? "" : " don't") <<
" need a runtime memory check.\n");
return CanVecMem;
// Check whether we found a reduction operator.
FoundReduxOp |= !IsAPhi;
- // Process users of current instruction. Push non PHI nodes after PHI nodes
+ // Process users of current instruction. Push non-PHI nodes after PHI nodes
// onto the stack. This way we are going to have seen all inputs to PHI
// nodes once we get to them.
SmallVector<Instruction *, 8> NonPHIs;
SmallVector<Instruction *, 8> PHIs;
- for (Value::use_iterator UI = Cur->use_begin(), E = Cur->use_end(); UI != E;
- ++UI) {
- Instruction *Usr = cast<Instruction>(*UI);
+ for (User *U : Cur->users()) {
+ Instruction *UI = cast<Instruction>(U);
// Check if we found the exit user.
- BasicBlock *Parent = Usr->getParent();
+ BasicBlock *Parent = UI->getParent();
if (!TheLoop->contains(Parent)) {
- // Exit if you find multiple outside users.
- if (ExitInstruction != 0)
+ // Exit if you find multiple outside users or if the header phi node is
+ // being used. In this case the user uses the value of the previous
+ // iteration, in which case we would loose "VF-1" iterations of the
+ // reduction operation if we vectorize.
+ if (ExitInstruction != 0 || Cur == Phi)
return false;
+
+ // The instruction used by an outside user must be the last instruction
+ // before we feed back to the reduction phi. Otherwise, we loose VF-1
+ // operations on the value.
+ if (std::find(Phi->op_begin(), Phi->op_end(), Cur) == Phi->op_end())
+ return false;
+
ExitInstruction = Cur;
continue;
}
- // Process instructions only once (termination).
- if (VisitedInsts.insert(Usr)) {
- if (isa<PHINode>(Usr))
- PHIs.push_back(Usr);
+ // Process instructions only once (termination). Each reduction cycle
+ // value must only be used once, except by phi nodes and min/max
+ // reductions which are represented as a cmp followed by a select.
+ ReductionInstDesc IgnoredVal(false, 0);
+ if (VisitedInsts.insert(UI)) {
+ if (isa<PHINode>(UI))
+ PHIs.push_back(UI);
else
- NonPHIs.push_back(Usr);
- }
+ NonPHIs.push_back(UI);
+ } else if (!isa<PHINode>(UI) &&
+ ((!isa<FCmpInst>(UI) &&
+ !isa<ICmpInst>(UI) &&
+ !isa<SelectInst>(UI)) ||
+ !isMinMaxSelectCmpPattern(UI, IgnoredVal).IsReduction))
+ return false;
+
// Remember that we completed the cycle.
- if (Usr == Phi)
+ if (UI == Phi)
FoundStartPHI = true;
}
Worklist.append(PHIs.begin(), PHIs.end());
// We must handle the select(cmp()) as a single instruction. Advance to the
// select.
if ((Cmp = dyn_cast<ICmpInst>(I)) || (Cmp = dyn_cast<FCmpInst>(I))) {
- if (!Cmp->hasOneUse() || !(Select = dyn_cast<SelectInst>(*I->use_begin())))
+ if (!Cmp->hasOneUse() || !(Select = dyn_cast<SelectInst>(*I->user_begin())))
return ReductionInstDesc(false, I);
return ReductionInstDesc(Select, Prev.MinMaxKind);
}
}
// We don't predicate stores at the moment.
- if (it->mayWriteToMemory() || it->mayThrow())
+ if (it->mayWriteToMemory()) {
+ StoreInst *SI = dyn_cast<StoreInst>(it);
+ // We only support predication of stores in basic blocks with one
+ // predecessor.
+ if (!SI || ++NumPredStores > NumberOfStoresToPredicate ||
+ !SafePtrs.count(SI->getPointerOperand()) ||
+ !SI->getParent()->getSinglePredecessor())
+ return false;
+ }
+ if (it->mayThrow())
return false;
+ // Check that we don't have a constant expression that can trap as operand.
+ for (Instruction::op_iterator OI = it->op_begin(), OE = it->op_end();
+ OI != OE; ++OI) {
+ if (Constant *C = dyn_cast<Constant>(*OI))
+ if (C->canTrap())
+ return false;
+ }
+
// The instructions below can trap.
switch (it->getOpcode()) {
default: continue;
return Factor;
}
+ if (!EnableCondStoresVectorization && Legal->NumPredStores) {
+ DEBUG(dbgs() << "LV: No vectorization. There are conditional stores.\n");
+ return Factor;
+ }
+
// Find the trip count.
unsigned TC = SE->getSmallConstantTripCount(TheLoop, TheLoop->getLoopLatch());
- DEBUG(dbgs() << "LV: Found trip count:"<<TC<<"\n");
+ DEBUG(dbgs() << "LV: Found trip count: " << TC << '\n');
unsigned WidestType = getWidestType();
unsigned WidestRegister = TTI.getRegisterBitWidth(true);
unsigned MaxSafeDepDist = -1U;
if (Legal->getMaxSafeDepDistBytes() != -1U)
MaxSafeDepDist = Legal->getMaxSafeDepDistBytes() * 8;
- WidestRegister = WidestRegister < MaxSafeDepDist ? WidestRegister : MaxSafeDepDist;
+ WidestRegister = ((WidestRegister < MaxSafeDepDist) ?
+ WidestRegister : MaxSafeDepDist);
unsigned MaxVectorSize = WidestRegister / WidestType;
DEBUG(dbgs() << "LV: The Widest type: " << WidestType << " bits.\n");
- DEBUG(dbgs() << "LV: The Widest register is:" << WidestRegister << "bits.\n");
+ DEBUG(dbgs() << "LV: The Widest register is: "
+ << WidestRegister << " bits.\n");
if (MaxVectorSize == 0) {
DEBUG(dbgs() << "LV: The target has no vector registers.\n");
if (UserVF != 0) {
assert(isPowerOf2_32(UserVF) && "VF needs to be a power of two");
- DEBUG(dbgs() << "LV: Using user VF "<<UserVF<<".\n");
+ DEBUG(dbgs() << "LV: Using user VF " << UserVF << ".\n");
Factor.Width = UserVF;
return Factor;
float Cost = expectedCost(1);
unsigned Width = 1;
- DEBUG(dbgs() << "LV: Scalar loop costs: "<< (int)Cost << ".\n");
+ 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: " <<
+ DEBUG(dbgs() << "LV: Vector loop of width " << i << " costs: " <<
(int)VectorCost << ".\n");
if (VectorCost < Cost) {
Cost = VectorCost;
}
}
- DEBUG(dbgs() << "LV: Selecting VF = : "<< Width << ".\n");
+ DEBUG(dbgs() << "LV: Selecting VF: "<< Width << ".\n");
Factor.Width = Width;
Factor.Cost = Width * Cost;
return Factor;
if (TC > 1 && TC < TinyTripCountUnrollThreshold)
return 1;
- unsigned TargetVectorRegisters = TTI.getNumberOfRegisters(true);
- DEBUG(dbgs() << "LV: The target has " << TargetVectorRegisters <<
- " vector registers\n");
+ unsigned TargetNumRegisters = TTI.getNumberOfRegisters(VF > 1);
+ DEBUG(dbgs() << "LV: The target has " << TargetNumRegisters <<
+ " registers\n");
+
+ if (VF == 1) {
+ if (ForceTargetNumScalarRegs.getNumOccurrences() > 0)
+ TargetNumRegisters = ForceTargetNumScalarRegs;
+ } else {
+ if (ForceTargetNumVectorRegs.getNumOccurrences() > 0)
+ TargetNumRegisters = ForceTargetNumVectorRegs;
+ }
LoopVectorizationCostModel::RegisterUsage R = calculateRegisterUsage();
// We divide by these constants so assume that we have at least one
// registers. These registers are used by all of the unrolled 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.
- unsigned UF = (TargetVectorRegisters - R.LoopInvariantRegs) / R.MaxLocalUsers;
+ // fit without causing spills. All of this is rounded down if necessary to be
+ // a power of two. We want power of two unroll factors to simplify any
+ // addressing operations or alignment considerations.
+ unsigned UF = PowerOf2Floor((TargetNumRegisters - R.LoopInvariantRegs) /
+ R.MaxLocalUsers);
+
+ // Don't count the induction variable as unrolled.
+ if (EnableIndVarRegisterHeur)
+ UF = PowerOf2Floor((TargetNumRegisters - R.LoopInvariantRegs - 1) /
+ std::max(1U, (R.MaxLocalUsers - 1)));
// Clamp the unroll factor ranges to reasonable factors.
unsigned MaxUnrollSize = TTI.getMaximumUnrollFactor();
+ // Check if the user has overridden the unroll max.
+ if (VF == 1) {
+ if (ForceTargetMaxScalarUnrollFactor.getNumOccurrences() > 0)
+ MaxUnrollSize = ForceTargetMaxScalarUnrollFactor;
+ } else {
+ if (ForceTargetMaxVectorUnrollFactor.getNumOccurrences() > 0)
+ MaxUnrollSize = ForceTargetMaxVectorUnrollFactor;
+ }
+
// 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)
else if (UF < 1)
UF = 1;
- if (Legal->getReductionVars()->size()) {
- DEBUG(dbgs() << "LV: Unrolling because of reductions. \n");
+ // Unroll if we vectorized this loop and there is a reduction that could
+ // benefit from unrolling.
+ if (VF > 1 && Legal->getReductionVars()->size()) {
+ DEBUG(dbgs() << "LV: Unrolling because of reductions.\n");
return UF;
}
- // We want to unroll tiny loops in order to reduce the loop overhead.
- // We assume that the cost overhead is 1 and we use the cost model
- // to estimate the cost of the loop and unroll until the cost of the
- // loop overhead is about 5% of the cost of the loop.
- DEBUG(dbgs() << "LV: Loop cost is "<< LoopCost <<" \n");
- if (LoopCost < 20) {
- DEBUG(dbgs() << "LV: Unrolling to reduce branch cost. \n");
- unsigned NewUF = 20/LoopCost + 1;
- return std::min(NewUF, UF);
+ // Note that if we've already vectorized the loop we will have done the
+ // runtime check and so unrolling won't require further checks.
+ bool UnrollingRequiresRuntimePointerCheck =
+ (VF == 1 && Legal->getRuntimePointerCheck()->Need);
+
+ // We want to unroll small loops in order to reduce the loop overhead and
+ // potentially expose ILP opportunities.
+ DEBUG(dbgs() << "LV: Loop cost is " << LoopCost << '\n');
+ if (!UnrollingRequiresRuntimePointerCheck &&
+ LoopCost < SmallLoopCost) {
+ // We assume that the cost overhead is 1 and we use the cost model
+ // to estimate the cost of the loop and unroll until the cost of the
+ // loop overhead is about 5% of the cost of the loop.
+ unsigned SmallUF = std::min(UF, (unsigned)PowerOf2Floor(SmallLoopCost / LoopCost));
+
+ // Unroll until store/load ports (estimated by max unroll factor) are
+ // saturated.
+ unsigned StoresUF = UF / (Legal->NumStores ? Legal->NumStores : 1);
+ unsigned LoadsUF = UF / (Legal->NumLoads ? Legal->NumLoads : 1);
+
+ if (EnableLoadStoreRuntimeUnroll && std::max(StoresUF, LoadsUF) > SmallUF) {
+ DEBUG(dbgs() << "LV: Unrolling to saturate store or load ports.\n");
+ return std::max(StoresUF, LoadsUF);
+ }
+
+ DEBUG(dbgs() << "LV: Unrolling to reduce branch cost.\n");
+ return SmallUF;
}
- DEBUG(dbgs() << "LV: Not Unrolling. \n");
+ DEBUG(dbgs() << "LV: Not Unrolling.\n");
return 1;
}
MaxUsage = std::max(MaxUsage, OpenIntervals.size());
DEBUG(dbgs() << "LV(REG): At #" << i << " Interval # " <<
- OpenIntervals.size() <<"\n");
+ OpenIntervals.size() << '\n');
// Add the current instruction to the list of open intervals.
OpenIntervals.insert(I);
}
unsigned Invariant = LoopInvariants.size();
- DEBUG(dbgs() << "LV(REG): Found max usage: " << MaxUsage << " \n");
- DEBUG(dbgs() << "LV(REG): Found invariant usage: " << Invariant << " \n");
- DEBUG(dbgs() << "LV(REG): LoopSize: " << R.NumInstructions << " \n");
+ DEBUG(dbgs() << "LV(REG): Found max usage: " << MaxUsage << '\n');
+ DEBUG(dbgs() << "LV(REG): Found invariant usage: " << Invariant << '\n');
+ DEBUG(dbgs() << "LV(REG): LoopSize: " << R.NumInstructions << '\n');
R.LoopInvariantRegs = Invariant;
R.MaxLocalUsers = MaxUsage;
continue;
unsigned C = getInstructionCost(it, VF);
- Cost += C;
- DEBUG(dbgs() << "LV: Found an estimated cost of "<< C <<" for VF " <<
- VF << " For instruction: "<< *it << "\n");
+
+ // Check if we should override the cost.
+ if (ForceTargetInstructionCost.getNumOccurrences() > 0)
+ C = ForceTargetInstructionCost;
+
+ BlockCost += C;
+ DEBUG(dbgs() << "LV: Found an estimated cost of " << C << " for VF " <<
+ VF << " For instruction: " << *it << '\n');
}
// We assume that if-converted blocks have a 50% chance of being executed.
// When the code is scalar then some of the blocks are avoided due to CF.
// When the code is vectorized we execute all code paths.
- if (Legal->blockNeedsPredication(*bb) && VF == 1)
+ if (VF == 1 && Legal->blockNeedsPredication(*bb))
BlockCost /= 2;
Cost += BlockCost;
return StepVal > MaxMergeDistance;
}
+static bool isStrideMul(Instruction *I, LoopVectorizationLegality *Legal) {
+ if (Legal->hasStride(I->getOperand(0)) || Legal->hasStride(I->getOperand(1)))
+ return true;
+ return false;
+}
+
unsigned
LoopVectorizationCostModel::getInstructionCost(Instruction *I, unsigned VF) {
// If we know that this instruction will remain uniform, check the cost of
case Instruction::And:
case Instruction::Or:
case Instruction::Xor: {
+ // Since we will replace the stride by 1 the multiplication should go away.
+ if (I->getOpcode() == Instruction::Mul && isStrideMul(I, Legal))
+ return 0;
// Certain instructions can be cheaper to vectorize if they have a constant
// second vector operand. One example of this are shifts on x86.
TargetTransformInfo::OperandValueKind Op1VK =
TargetTransformInfo::OK_AnyValue;
TargetTransformInfo::OperandValueKind Op2VK =
TargetTransformInfo::OK_AnyValue;
+ Value *Op2 = I->getOperand(1);
- if (isa<ConstantInt>(I->getOperand(1)))
+ // Check for a splat of a constant or for a non uniform vector of constants.
+ if (isa<ConstantInt>(Op2))
Op2VK = TargetTransformInfo::OK_UniformConstantValue;
+ else if (isa<ConstantVector>(Op2) || isa<ConstantDataVector>(Op2)) {
+ Op2VK = TargetTransformInfo::OK_NonUniformConstantValue;
+ if (cast<Constant>(Op2)->getSplatValue() != NULL)
+ Op2VK = TargetTransformInfo::OK_UniformConstantValue;
+ }
return TTI.getArithmeticInstrCost(I->getOpcode(), VectorTy, Op1VK, Op2VK);
}
static const char lv_name[] = "Loop Vectorization";
INITIALIZE_PASS_BEGIN(LoopVectorize, LV_NAME, lv_name, false, false)
INITIALIZE_AG_DEPENDENCY(TargetTransformInfo)
+INITIALIZE_PASS_DEPENDENCY(BlockFrequencyInfo)
+INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
INITIALIZE_PASS_DEPENDENCY(ScalarEvolution)
+INITIALIZE_PASS_DEPENDENCY(LCSSA)
+INITIALIZE_PASS_DEPENDENCY(LoopInfo)
INITIALIZE_PASS_DEPENDENCY(LoopSimplify)
INITIALIZE_PASS_END(LoopVectorize, LV_NAME, lv_name, false, false)
namespace llvm {
- Pass *createLoopVectorizePass() {
- return new LoopVectorize();
+ Pass *createLoopVectorizePass(bool NoUnrolling, bool AlwaysVectorize) {
+ return new LoopVectorize(NoUnrolling, AlwaysVectorize);
}
}
return false;
}
+
+
+void InnerLoopUnroller::scalarizeInstruction(Instruction *Instr,
+ bool IfPredicateStore) {
+ assert(!Instr->getType()->isAggregateType() && "Can't handle vectors");
+ // Holds vector parameters or scalars, in case of uniform vals.
+ SmallVector<VectorParts, 4> Params;
+
+ setDebugLocFromInst(Builder, Instr);
+
+ // 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(getVectorValue(SrcOp));
+ 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 && OrigLoop->contains(SrcInst)) {
+ assert(WidenMap.has(SrcInst) && "Source operand is unavailable");
+ // The parameter is a vector value from earlier.
+ Params.push_back(WidenMap.get(SrcInst));
+ } else {
+ // The parameter is a scalar from outside the loop. Maybe even a constant.
+ VectorParts Scalars;
+ Scalars.append(UF, SrcOp);
+ Params.push_back(Scalars);
+ }
+ }
+
+ assert(Params.size() == Instr->getNumOperands() &&
+ "Invalid number of operands");
+
+ // Does this instruction return a value ?
+ bool IsVoidRetTy = Instr->getType()->isVoidTy();
+
+ Value *UndefVec = IsVoidRetTy ? 0 :
+ UndefValue::get(Instr->getType());
+ // Create a new entry in the WidenMap and initialize it to Undef or Null.
+ VectorParts &VecResults = WidenMap.splat(Instr, UndefVec);
+
+ Instruction *InsertPt = Builder.GetInsertPoint();
+ BasicBlock *IfBlock = Builder.GetInsertBlock();
+ BasicBlock *CondBlock = 0;
+
+ VectorParts Cond;
+ Loop *VectorLp = 0;
+ if (IfPredicateStore) {
+ assert(Instr->getParent()->getSinglePredecessor() &&
+ "Only support single predecessor blocks");
+ Cond = createEdgeMask(Instr->getParent()->getSinglePredecessor(),
+ Instr->getParent());
+ VectorLp = LI->getLoopFor(IfBlock);
+ assert(VectorLp && "Must have a loop for this block");
+ }
+
+ // For each vector unroll 'part':
+ for (unsigned Part = 0; Part < UF; ++Part) {
+ // For each scalar that we create:
+
+ // Start an "if (pred) a[i] = ..." block.
+ Value *Cmp = 0;
+ if (IfPredicateStore) {
+ if (Cond[Part]->getType()->isVectorTy())
+ Cond[Part] =
+ Builder.CreateExtractElement(Cond[Part], Builder.getInt32(0));
+ Cmp = Builder.CreateICmp(ICmpInst::ICMP_EQ, Cond[Part],
+ ConstantInt::get(Cond[Part]->getType(), 1));
+ CondBlock = IfBlock->splitBasicBlock(InsertPt, "cond.store");
+ LoopVectorBody.push_back(CondBlock);
+ VectorLp->addBasicBlockToLoop(CondBlock, LI->getBase());
+ // Update Builder with newly created basic block.
+ Builder.SetInsertPoint(InsertPt);
+ }
+
+ Instruction *Cloned = Instr->clone();
+ if (!IsVoidRetTy)
+ Cloned->setName(Instr->getName() + ".cloned");
+ // Replace the operands of the cloned instructions with extracted scalars.
+ for (unsigned op = 0, e = Instr->getNumOperands(); op != e; ++op) {
+ Value *Op = Params[op][Part];
+ 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[Part] = Cloned;
+
+ // End if-block.
+ if (IfPredicateStore) {
+ BasicBlock *NewIfBlock = CondBlock->splitBasicBlock(InsertPt, "else");
+ LoopVectorBody.push_back(NewIfBlock);
+ VectorLp->addBasicBlockToLoop(NewIfBlock, LI->getBase());
+ Builder.SetInsertPoint(InsertPt);
+ Instruction *OldBr = IfBlock->getTerminator();
+ BranchInst::Create(CondBlock, NewIfBlock, Cmp, OldBr);
+ OldBr->eraseFromParent();
+ IfBlock = NewIfBlock;
+ }
+ }
+}
+
+void InnerLoopUnroller::vectorizeMemoryInstruction(Instruction *Instr) {
+ StoreInst *SI = dyn_cast<StoreInst>(Instr);
+ bool IfPredicateStore = (SI && Legal->blockNeedsPredication(SI->getParent()));
+
+ return scalarizeInstruction(Instr, IfPredicateStore);
+}
+
+Value *InnerLoopUnroller::reverseVector(Value *Vec) {
+ return Vec;
+}
+
+Value *InnerLoopUnroller::getBroadcastInstrs(Value *V) {
+ return V;
+}
+
+Value *InnerLoopUnroller::getConsecutiveVector(Value* Val, int StartIdx,
+ bool Negate) {
+ // When unrolling and the VF is 1, we only need to add a simple scalar.
+ Type *ITy = Val->getType();
+ assert(!ITy->isVectorTy() && "Val must be a scalar");
+ Constant *C = ConstantInt::get(ITy, StartIdx, Negate);
+ return Builder.CreateAdd(Val, C, "induction");
+}
+