[all-commits] [llvm/llvm-project] 467624: AMDGPU/GFX12: Do not wait unnecessarily before bar...

[Nicolai Hähnle via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 46762421c30a361c439ad5930f1fd026601db7f5
      https://github.com/llvm/llvm-project/commit/46762421c30a361c439ad5930f1fd026601db7f5
  Author: Nicolai Hähnle <nicolai.haehnle at amd.com>
  Date:   2025-08-23 (Sat, 23 Aug 2025)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
    M llvm/lib/Target/AMDGPU/SIInstrInfo.h
    M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.barrier.ll
    M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.barrier.signal.isfirst.ll
    M llvm/test/CodeGen/AMDGPU/s-barrier.ll

  Log Message:
  -----------
  AMDGPU/GFX12: Do not wait unnecessarily before barriers (#154970)

The barrier intrinsic itself should not have memory semantics. Frontends
should use appropriate fence instructions for memory effects, and some
frontends want to rely on that for performance (e.g. wait only for LDS
before a barrier).

See the code comment for more detail.



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