[all-commits] [llvm/llvm-project] 92083e: [X86] Allow VPERMV3 -> VPERMV folds to handle extr...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 92083e855b4d4ce7e3f7633cd35a4fcb90e2c24f
      https://github.com/llvm/llvm-project/commit/92083e855b4d4ce7e3f7633cd35a4fcb90e2c24f
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2024-07-08 (Mon, 08 Jul 2024)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/pr97968.ll

  Log Message:
  -----------
  [X86] Allow VPERMV3 -> VPERMV folds to handle extraction from a wider source vector (e.g. v16i32 -> v4i32)

We don't need to restrict this to double width vectors, as long as we correctly bitcast the types

Improves the fix for #97968



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