[PATCH] D104624: [LoopVectorize] Add support for scalable vectorization of invariant stores

Peter Waller via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Jun 28 08:32:30 PDT 2021


peterwaller-arm accepted this revision.
peterwaller-arm added a comment.
This revision is now accepted and ready to land.

I'm relatively inexperienced in this area of the code, but giving this a +1 on the basis of reading the diff and thinking about this.

I understand that the result might not be ideal because a scatter may be slow when every lane is writing to the same address. I also understand that it seems reasonable to do this in the short term to reach functional correctness.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D104624/new/

https://reviews.llvm.org/D104624



More information about the llvm-commits mailing list