-
Notifications
You must be signed in to change notification settings - Fork 22.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[c10d] add fake pg necessary collectives #102238
Conversation
This PR adds fake pg necessary collectives to enable e2e FSDP run with out multiprocess or multithreading [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/102238
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit a21ea0a: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This PR adds fake pg necessary collectives to enable e2e FSDP run with out multiprocess or multithreading ghstack-source-id: 13e76fc541acab17e5a12a933f6ba033646bc06a Pull Request resolved: #102238
# Fake process group doesn't need barrier | ||
pass | ||
elif backend == Backend.MPI: | ||
if backend == Backend.MPI: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nyehh, I think I preferred not mucking about with the environment variable instead haha.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah I kinda feel bad for this too, it was easier for me to leverage the previous work that uses a environ knob to control barrier init, let me think about it, there should be some better way for custom backend.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
landing this first to unblock dev efficiency and will address this in a follow up PR
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
follow up PR here: #102343
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Stack from ghstack (oldest at bottom):
This PR adds fake pg necessary collectives to enable e2e FSDP run
with out multiprocess or multithreading