[PATCH] D92701: [SLPVectorize] Call isLegalMaskedGather before creating a gather TreeEntry

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added a comment.

Yes improving the cost model makes more sense (both for gather/scatter and gep vectorization costs) - SLP should only ever create a gather with a constant mask, so at the very least ScalarizeMaskedMemIntrin should do a good job of converting the loads to a BUILD_VECTOR sequence.

If we can get examples of bad vectorization using gathers I can take a look at working out exactly where the cost model is falling down.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D92701/new/

https://reviews.llvm.org/D92701