[PATCH] D66071: [X86] Teach lowerV4I32Shuffle to only use broadcasts if the mask has more than one undef element. Prioritize shifts over broadcast in lowerV8I16Shuffle.

Mon Aug 19 05:53:05 PDT 2019

RKSimon accepted this revision.
RKSimon added a comment.
This revision is now accepted and ready to land.

In D66071#1625242 <https://reviews.llvm.org/D66071#1625242>, @craig.topper wrote:

> Are you asking about v8i16 where I switched the shift priority?

Yes, but I think combineX86ShuffleChain will try to convert this to a foldableload+broadcast if it can so we should be OK.

LGTM

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66071/new/

https://reviews.llvm.org/D66071