Part 1: Introduce multi-node SPMD support for Neuron #8204
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
rpsilva-aws
changed the title
|
Note, same as: #8046 (review) |
rpsilva-aws
changed the title
|
@will-cromar, fyi: #8204 (comment) |
JackCaoG
reviewed
Oct 2, 2024
rpsilva-aws
force-pushed
the
rpsilva-aws_neuron_multi_node_spmd_pt25
branch
from
7934779
to
f221f59
Compare
JackCaoG
approved these changes
Oct 2, 2024
|
please fix the linter |
rpsilva-aws
force-pushed
the
rpsilva-aws_neuron_multi_node_spmd_pt25
branch
from
f221f59
to
0571c47
Compare
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Outdated suggestions cannot be applied.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.
In this PR, we adapt to account for a new initialization path that supports multi-node SPMD in Neuron. In order to minimize this change, we retain the xla.init() API, but introduce a reinitialization for PJRT alone once SPMD is enabled. Since enabling SPMD follows the initial Neuron initialization, we require reconfiguring once this is enabled, and if the user did not explicitly set XLA_USE_SPMD (via is_spmd(), as it is currently recommended). Under the hood, both APIs will guarantee that the environment is correctly configured when SPMD is enabled.
In a follow-up, the reinitialization path is moved to torch-xla.