[all-commits] [llvm/llvm-project] f30f7a: [X86] canonicalizeShuffleWithOp - initial support ...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: f30f7a084c93bb7ef4983cda583985b56ad123d2
      https://github.com/llvm/llvm-project/commit/f30f7a084c93bb7ef4983cda583985b56ad123d2
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2024-11-27 (Wed, 27 Nov 2024)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/vector-half-conversions.ll

  Log Message:
  -----------
  [X86] canonicalizeShuffleWithOp - initial support for shuffle(cvt(x),cvt(y)) -> cvt(shuffle(x,y))

Initial support is just for UNPCKL(CVTPH2PS(X),CVTPH2PS(Y)) -> CVTPH2PS(UNPCKL(X,Y))

Making this more general for other shuffles/conversions will have to be done carefully as we have to handle changes in src/dst element width, so I just handled the CVTPH2PS regression case.

Fixes #83414



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