[all-commits] [llvm/llvm-project] db5f24: AMDGPU: Handle V->A MFMA copy from case with immed...

[Matt Arsenault via All-commits]

  Branch: refs/heads/users/arsenm/amdgpu/handle-subreg-def-read-mfma-copy-from-agpr
  Home:   https://github.com/llvm/llvm-project
  Commit: db5f240cb9df69b065f96a018e1281c1eff93d4a
      https://github.com/llvm/llvm-project/commit/db5f240cb9df69b065f96a018e1281c1eff93d4a
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2025-08-21 (Thu, 21 Aug 2025)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPURewriteAGPRCopyMFMA.cpp
    M llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-copy-from.mir

  Log Message:
  -----------
  AMDGPU: Handle V->A MFMA copy from case with immediate src2

Handle a special case for copies from AGPR VGPR on the MFMA inputs.
If the "input" is really a subregister def, we will not see the
usual copy to VGPR for src2, only the read of the subregister def.
Not sure if this pattern appears in practice.



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