[all-commits] [llvm/llvm-project] cfd1d4: [AMDGPU] Relax lds dma waitcnt with no aliasing pair

[Austin Kerbow via All-commits]

  Branch: refs/heads/users/kerbowa/rexlax-lds-dma-wait
  Home:   https://github.com/llvm/llvm-project
  Commit: cfd1d488e05296fa4651b7bd5a880fb56fd2bdec
      https://github.com/llvm/llvm-project/commit/cfd1d488e05296fa4651b7bd5a880fb56fd2bdec
  Author: Austin Kerbow <Austin.Kerbow at amd.com>
  Date:   2025-03-21 (Fri, 21 Mar 2025)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
    M llvm/test/CodeGen/AMDGPU/lds-dma-waits.ll

  Log Message:
  -----------
  [AMDGPU] Relax lds dma waitcnt with no aliasing pair

If we cannot find any lds DMA instruction that is aliased by some load
from lds, we will still insert vmcnt(0). This is overly cautious since
handling inter-thread dependences is normally managed by the memory
model instead of the waitcnt pass, so this change updates the behavior
to be more inline with how other types of memory events are handled.



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