TargetTransformInfo: address calculation parameter for gather/scather

Address calculation for gather/scather in vectorized code can incur a
significant cost making vectorization unbeneficial. Add infrastructure to add
cost.
Tests and cost model for targets will be in follow-up commits.

radar://14351991

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186187 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 56 insertions, 1 deletions

diff --git a/lib/Transforms/Vectorize/LoopVectorize.cpp b/lib/Transforms/Vectorize/LoopVectorize.cpp
index 020eb61..b35ed74 100644
--- a/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -4403,6 +4403,59 @@ unsigned LoopVectorizationCostModel::expectedCost(unsigned VF) {
   return Cost;
 }
 
+/// \brief Check whether the address computation for a non-consecutive memory
+/// access looks like an unlikely candidate for being merged into the indexing
+/// mode.
+///
+/// We look for a GEP which has one index that is an induction variable and all
+/// other indices are loop invariant. If the stride of this access is also
+/// within a small bound we decide that this address computation can likely be
+/// merged into the addressing mode.
+/// In all other cases, we identify the address computation as complex.
+static bool isLikelyComplexAddressComputation(Value *Ptr,
+                                              LoopVectorizationLegality *Legal,
+                                              ScalarEvolution *SE,
+                                              const Loop *TheLoop) {
+  GetElementPtrInst *Gep = dyn_cast<GetElementPtrInst>(Ptr);
+  if (!Gep)
+    return true;
+
+  // We are looking for a gep with all loop invariant indices except for one
+  // which should be an induction variable.
+  unsigned NumOperands = Gep->getNumOperands();
+  for (unsigned i = 1; i < NumOperands; ++i) {
+    Value *Opd = Gep->getOperand(i);
+    if (!SE->isLoopInvariant(SE->getSCEV(Opd), TheLoop) &&
+        !Legal->isInductionVariable(Opd))
+      return true;
+  }
+
+  // Now we know we have a GEP ptr, %inv, %ind, %inv. Make sure that the step
+  // can likely be merged into the address computation.
+  unsigned MaxMergeDistance = 64;
+
+  const SCEVAddRecExpr *AddRec = dyn_cast<SCEVAddRecExpr>(SE->getSCEV(Ptr));
+  if (!AddRec)
+    return true;
+
+  // Check the step is constant.
+  const SCEV *Step = AddRec->getStepRecurrence(*SE);
+  // Calculate the pointer stride and check if it is consecutive.
+  const SCEVConstant *C = dyn_cast<SCEVConstant>(Step);
+  if (!C)
+    return true;
+
+  const APInt &APStepVal = C->getValue()->getValue();
+
+  // Huge step value - give up.
+  if (APStepVal.getBitWidth() > 64)
+    return true;
+
+  int64_t StepVal = APStepVal.getSExtValue();
+
+  return StepVal > MaxMergeDistance;
+}
+
 unsigned
 LoopVectorizationCostModel::getInstructionCost(Instruction *I, unsigned VF) {
   // If we know that this instruction will remain uniform, check the cost of
@@ -4498,6 +4551,8 @@ LoopVectorizationCostModel::getInstructionCost(Instruction *I, unsigned VF) {
     unsigned ScalarAllocatedSize = DL->getTypeAllocSize(ValTy);
     unsigned VectorElementSize = DL->getTypeStoreSize(VectorTy)/VF;
     if (!ConsecutiveStride || ScalarAllocatedSize != VectorElementSize) {
+      bool IsComplexComputation =
+        isLikelyComplexAddressComputation(Ptr, Legal, SE, TheLoop);
       unsigned Cost = 0;
       // The cost of extracting from the value vector and pointer vector.
       Type *PtrTy = ToVectorTy(Ptr->getType(), VF);
@@ -4513,7 +4568,7 @@ LoopVectorizationCostModel::getInstructionCost(Instruction *I, unsigned VF) {
       }
 
       // The cost of the scalar loads/stores.
-      Cost += VF * TTI.getAddressComputationCost(ValTy->getScalarType());
+      Cost += VF * TTI.getAddressComputationCost(PtrTy, IsComplexComputation);
       Cost += VF * TTI.getMemoryOpCost(I->getOpcode(), ValTy->getScalarType(),
                                        Alignment, AS);
       return Cost;