[PATCH] D103818: [X86] lowerVECTOR_SHUFFLE(): allow widening shuffle to have i128 (YMM half) element type

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added a comment.

Why is this better? vinsertf128 tends to be faster than broadcasts



================
Comment at: llvm/test/CodeGen/X86/vector-shuffle-256-v4.ll:1016
 ; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
 ; AVX2-NEXT:    retq
 ;
----------------
The AVX1 shuffle looks to be much better............


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D103818/new/

https://reviews.llvm.org/D103818