Skip to content

Commit 0ba3e39

Browse files
authored
[CI] Spawn docker container with 2Gb shmem
Should prevent crashes during NCCL initialization. If `data_parallel_tutorial.py` is executed without this option it would segfault in `ncclShmOpen` while executing ` nn.DataParallel(model)`
1 parent f0e587e commit 0ba3e39

File tree

1 file changed

+1
-0
lines changed

1 file changed

+1
-0
lines changed

.github/workflows/build-tutorials.yml

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -87,6 +87,7 @@ jobs:
8787
--tty \
8888
--detach \
8989
--user jenkins \
90+
--shm-size=2gb \
9091
--name="${container_name}" \
9192
-v "${GITHUB_WORKSPACE}:/var/lib/jenkins/workspace" \
9293
-w /var/lib/jenkins/workspace \

0 commit comments

Comments
 (0)