[Feat] custom ring all-reduce #23
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
* add all reduce test * introduce different all reduce backend and add testing * add custom backend to all reduce test * fix group passing gloo al reduce * use cap because jack ask for it even tho its ugly
samsja
reviewed
Oct 3, 2024
samsja
reviewed
Oct 3, 2024
samsja
reviewed
Oct 3, 2024
samsja
reviewed
Oct 4, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
torchrun --nproc-per-node 2 scripts/all_reduce_test.py --backend custom --transfer_dtype uint8onsaturnHowever under the 500mbps test, it beats the gloo one
1B all reduce takes 102s under 500mbps conditions (I think the 4 procs share the limit tho so maybe we can divide the time taken by 4 if the links are all individually 500mbps)
