[PATCH] D125238: [X86] Prefer MOVHLPS for shuffle(x, 1, -1) extraction patterns (PR26515)

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added inline comments.


================
Comment at: llvm/test/CodeGen/X86/vector-reduce-fadd.ll:54
+; SSE2-NEXT:    xorps %xmm2, %xmm2
+; SSE2-NEXT:    movhlps {{.*#+}} xmm2 = xmm1[1],xmm2[1]
 ; SSE2-NEXT:    addss %xmm2, %xmm0
----------------
TBH I'm not convinced any of these need dependency breaks.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D125238/new/

https://reviews.llvm.org/D125238