[PATCH] D48963: [X86][SSE] Combine v16i8 SHL by constants to multiplies

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added a comment.

In https://reviews.llvm.org/D48963#1153375, @craig.topper wrote:

> Should v32i8 do the same or is there something different about it?


Some cases AVX2 gives smaller code with the PBLENDVB generic shift pattern - but I didn't spend that much time looking at it - I'll take another look.


Repository:
  rL LLVM

https://reviews.llvm.org/D48963