[all-commits] [llvm/llvm-project] fbfd78: [X86] lowerShuffleAsRepeatedMaskAndLanePermute - a...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: fbfd78f7aae520421adf4e1f7b35a48a3fb5a8be
      https://github.com/llvm/llvm-project/commit/fbfd78f7aae520421adf4e1f7b35a48a3fb5a8be
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2022-04-03 (Sun, 03 Apr 2022)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/vector-interleaved-load-i8-stride-4.ll
    M llvm/test/CodeGen/X86/vector-shuffle-512-v64.ll
    M llvm/test/CodeGen/X86/x86-interleaved-access.ll

  Log Message:
  -----------
  [X86] lowerShuffleAsRepeatedMaskAndLanePermute - allow v16i32 sub-lane permutes for v64i8 shuffles

Without VBMI, we are better off permuting v16i32 sub-lanes, even though its a variable shuffle, if it allows us to then shuffle v64i8 inlane repeated masks (PSHUFB etc.)

Fixes #54658