[all-commits] [llvm/llvm-project] 5777c7: [SLP][AMDGPU] Fix load-address typo in v4f16 combi...

[Pablo Reble via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 5777c75fd91efa7abce23a74c4ec2381e8084b44
      https://github.com/llvm/llvm-project/commit/5777c75fd91efa7abce23a74c4ec2381e8084b44
  Author: Pablo Reble <pablo.reble at amd.com>
  Date:   2026-09-10 (Thu, 10 Sep 2026)

  Changed paths:
    M llvm/test/Transforms/SLPVectorizer/AMDGPU/slp-v2f16.ll

  Log Message:
  -----------
  [SLP][AMDGPU] Fix load-address typo in v4f16 combine tests (#222112)

Patch addresses issue where tests are not performing 4 contiguous
elements load/store chain as intended.

- Correcting both typos (`%tmp6` vs. `%tmp10`) so each function
represents a contiguous 4-element chain.
- `copysign_combine_v4f16` and `canonicalize_combine_v4f16` now
vectorize identically on `GFX8` and `GFX9`.
- Adding `canonicalize_combine_v4f16_noncontiguous`, to preserve
existing test coverage, so the non-contiguous-gather path isn't silently
dropped.



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