[all-commits] [llvm/llvm-project] a728d6: AMDGPU: Add test showing bit operations that shoul...

[Matt Arsenault via All-commits]

  Branch: refs/heads/users/arsenm/add-not-constant-0-test
  Home:   https://github.com/llvm/llvm-project
  Commit: a728d63ac0c9d78a949cc608a614bc2da8733bab
      https://github.com/llvm/llvm-project/commit/a728d63ac0c9d78a949cc608a614bc2da8733bab
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2025-06-10 (Tue, 10 Jun 2025)

  Changed paths:
    A llvm/test/CodeGen/AMDGPU/bit-op-reduce-width-known-bits.ll

  Log Message:
  -----------
  AMDGPU: Add test showing bit operations that should be reducible

v_xor_i64_known_i32_from_range_use_out_of_block demonstrates a regression
that appears in a future patch in the IR division expansion.

We could generalize splitBinaryBitConstantOp to use known bits. I'm
not sure if it's worth it in the original example, since the pattern
seems to disappear if I optimize the division expansion. We should
probably fix the divide expand to avoid this.



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