[PATCH] D81085: [AMDGPU/MemOpsCluster] Implement new heuristic for computing max mem ops cluster size

[Mahesha S via Phabricator via llvm-commits]

hsmhsm added a comment.

Hi @thakis

This is surprizing,  in my local repo the tests passes, but, it fails in pre-checkin build. I have locally built the patch from scratch with upstream rebased repo, and tested it as below:

  mahesha at brego:[AMDGPU]$ ~/ROCm/github/llvm/rel-build/bin/llvm-lit -vv cluster_stores.ll llvm.amdgcn.image.nsa.ll unigine-liveness-crash.ll wqm.ll
  -- Testing: 4 tests, 4 workers --
  UNSUPPORTED: LLVM :: CodeGen/AMDGPU/cluster_stores.ll (1 of 4)
  PASS: LLVM :: CodeGen/AMDGPU/unigine-liveness-crash.ll (2 of 4)
  PASS: LLVM :: CodeGen/AMDGPU/llvm.amdgcn.image.nsa.ll (3 of 4)
  PASS: LLVM :: CodeGen/AMDGPU/wqm.ll (4 of 4)
  
  Testing Time: 0.31s
    Unsupported: 1
    Passed     : 3
  mahesha at brego:[AMDGPU]$


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D81085/new/

https://reviews.llvm.org/D81085