[all-commits] [llvm/llvm-project] 718d50: [VectorCombine] foldPermuteOfBinops - prefer the n...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 718d50d6d03449962b14a3c8357a6ee3fa145f36
      https://github.com/llvm/llvm-project/commit/718d50d6d03449962b14a3c8357a6ee3fa145f36
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2024-11-01 (Fri, 01 Nov 2024)

  Changed paths:
    M llvm/lib/Transforms/Vectorize/VectorCombine.cpp
    M llvm/test/Transforms/PhaseOrdering/X86/pr50392.ll
    M llvm/test/Transforms/PhaseOrdering/X86/pr94546.ll

  Log Message:
  -----------
  [VectorCombine] foldPermuteOfBinops - prefer the new fold for matching costs.

Minor tweak to #114101 - as we're reducing the instruction count, we should prefer the fold if the old/new costs are the same.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications