OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-07 12:21:27 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity

[CI] Do not constrain memory for ROCm testing in CI (#156115)

Browse Source 
Fixes ROCm OOMs introduced by https://github.com/pytorch/pytorch/pull/155631

Pull Request resolved: https://github.com/pytorch/pytorch/pull/156115
Approved by: https://github.com/jeffdaily
This commit is contained in:
Catherine Lee 2025-06-17 15:30:36 +00:00 committed by PyTorch MergeBot
parent 7fcad0231c
commit 0079c80b35
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
test/run_test.py
View File

@ -1855,7 +1855,9 @@ def run_tests(
): ):
raise RuntimeError(failure.message + keep_going_message) raise RuntimeError(failure.message + keep_going_message)
os.environ["NUM_PARALLEL_PROCS"] = str(NUM_PROCS) # This is used later to constrain memory per proc on the GPU. On ROCm
# the number of procs is the number of GPUs, so we don't need to do this
os.environ["NUM_PARALLEL_PROCS"] = str(1 if torch.version.hip else NUM_PROCS)
# See Note [ROCm parallel CI testing] # See Note [ROCm parallel CI testing]
pool = get_context("spawn").Pool( pool = get_context("spawn").Pool(