[all-commits] [llvm/llvm-project] 639d7c: AMDGPU: Refine gfx950 xdl-write-vgpr hazard cases

Matt Arsenault via All-commits all-commits at lists.llvm.org
Fri Nov 22 20:32:11 PST 2024


  Branch: refs/heads/users/arsenm/gfx950/refine-xdl-write-vgpr-hazard-cases
  Home:   https://github.com/llvm/llvm-project
  Commit: 639d7c62daa3adf6a054638b18936672ab0bffc3
      https://github.com/llvm/llvm-project/commit/639d7c62daa3adf6a054638b18936672ab0bffc3
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2024-11-23 (Sat, 23 Nov 2024)

  Changed paths:
    M llvm/lib/Target/AMDGPU/GCNHazardRecognizer.cpp
    M llvm/test/CodeGen/AMDGPU/mai-hazards-gfx940.mir

  Log Message:
  -----------
  AMDGPU: Refine gfx950 xdl-write-vgpr hazard cases

The 2-pass XDL write VGPR, read by non-XDL SGEMM/DGEMM case
was 1 wait state overly conservative. Previously, for gfx940,
the XDL/non-XDL cases happened to have the same number of cycles
in all cases. Now the XDL consumer case has an additional state for
2 pass sources.



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


More information about the All-commits mailing list