[PATCH] D112552: [LoopVectorize] When tail-folding, don't always predicate uniform loads

[David Sherwood via Phabricator via llvm-commits]

david-arm added inline comments.


================
Comment at: llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:9070
+  if (IsUniform && isa<LoadInst>(I) &&
+      !Legal->blockNeedsPredication(I->getParent()) && Range.Start.isVector()) {
+    assert(CM.foldTailByMasking() &&
----------------
fhahn wrote:
> The problem with the workaround is that `IsPredicated` is now inaccurate for the recipe I think.
> 
> It looks like the cost-model already accurately estimates the cost as not requiring predication. So perhaps it would be better to not claim the instruction requires predication in the first place in `LoopVectorizationCostModel::isPredicated`?
Hi @fhahn, are you suggesting that `IsPredicated` is being set incorrectly above? If I understand correctly, I believe you're suggesting the answer would be to add more logic to `isPredicatedInst`? i.e. something like:

  bool isPredicatedInst(Instruction *I, bool IsUniform) {
    if (IsUniform && isa<LoadInst>(I) &&
      !Legal->blockNeedsPredication(I->getParent()))
      return false;

    if (!blockNeedsPredication(I->getParent()))
    ...
  }

That feels a bit messy. Would it be better to simply adjust the IsPredicated boolean value instead in this function?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D112552/new/

https://reviews.llvm.org/D112552