Support distributed init outside runtime init #276
Closed
Conversation
This file contains hidden or 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
tjruwase
approved these changes
Jun 30, 2020
| self.device = torch.device("cuda", self.local_rank) | ||
| self.world_size = dist.get_world_size() | ||
| self.global_rank = dist.get_rank() | ||
| logger.info("Set device to local rank {} within node.".format( |
Contributor
There was a problem hiding this comment.
Can we move this message to where we set self.local_rank in line 125?
jeffra
requested review from
RezaYazdaniAminabadi,
arashashari,
awan-10,
cli99,
conglongli,
eltonzheng,
minjiaz and
niumanar
as code owners
andrewchoi5
approved these changes
Oct 19, 2020
jeffra
added a commit
that referenced
this pull request
Dec 17, 2020
Collaborator
Author
|
Closing and moving to #608 |
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Outdated suggestions cannot be applied.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.
Allow users to initialize torch distributed outside of deepspeed.initialize. Specifically needed for MPI discovery (e.g., AML) and other models. This also removes the need for --deepspeed_mpi flag, we will automatically attempt to discover MPI world info if we are not launched with a torch.distributed or deepspeed launcher.
TODO: Need to update documentation, remove --deepspeed_mpi flag, and add unit tests