[all-commits] [llvm/llvm-project] 5636eb: [X86] combineBlendOfPermutes - allow whole-lane pe...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 5636eb89bd69f9c55f4e4aeafaa8c04aa99e5c84
      https://github.com/llvm/llvm-project/commit/5636eb89bd69f9c55f4e4aeafaa8c04aa99e5c84
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2024-05-08 (Wed, 08 May 2024)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/vector-shuffle-combining-avx.ll

  Log Message:
  -----------
  [X86] combineBlendOfPermutes - allow whole-lane permutation on AVX1 targets.

dd4bf22b9380e797362fac1415a1796da338b2db fixed #91433 but meant we couldn't use vperm2f128 to permute entire 128-bit lanes - if the new 256-bit permutation mask can be scaled to 2x128-bit elements, then we can still fold.



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