[all-commits] [llvm/llvm-project] 55e8c5: [AMDGPU] Mark ASYNCMARK as meta instruction to fix...

[adeshcom14 via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 55e8c5e0eaea9943c2039a8add9525ed90dad03a
      https://github.com/llvm/llvm-project/commit/55e8c5e0eaea9943c2039a8add9525ed90dad03a
  Author: adeshcom14 <aadikane at amd.com>
  Date:   2026-04-17 (Fri, 17 Apr 2026)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
    M llvm/lib/Target/AMDGPU/SOPInstructions.td
    M llvm/test/CodeGen/AMDGPU/async-buffer-loads.ll
    M llvm/test/CodeGen/AMDGPU/asyncmark-gfx12plus.ll
    M llvm/test/CodeGen/AMDGPU/asyncmark-pregfx12.ll

  Log Message:
  -----------
  [AMDGPU] Mark ASYNCMARK as meta instruction to fix hazard cycle miscounting (#189981)

ASYNCMARK emits no hardware code it is used for tracking purpose but was
not marked as meta, causing getNumWaitStates to return 1 and
GCNHazardRecognizer to incorrectly count it as a pipeline cycle.
This patch marks ASYNCMARK as meta-Instruction so it correctly reports 0
wait states.

Fixes: #186878



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