PR #18254: [XLA:GPU] Simplifies All-Reduce if they can be simplified or are degenerated. #18297
+76
−0
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
![copybara-service[bot]](https://avatars.githubusercontent.com/in/44061?s=60&v=4){kind=small}
copybara-service
bot
force-pushed
the
test_685835366
branch
11 times, most recently
from
84d548b
to
8d42868
Compare
…or are degenerated. Imported from GitHub PR #18254 This PR adds back the AllReduceSimplifier pass to the GPU compiler as in some cases we see JAX programs will add some degenerated psum to the HLOs, Code pointer in JAX: 1) https://cs.opensource.google/jax/jax/+/main:jax/experimental/shard_map.py;drc=875f44c63a6c0f082ef3b228559e86e1d0a398d1;l=1674 2) pbroadcast to psum conversions. The compiler should generally remove these unnecessary ARs so that it unblocks many optimizations like collective pipeliner. It also emits unncessary runtime thunks as well. Copybara import of the project: -- b89df55 by Yunlong Liu <yunlongl@x.ai>: Removes degenerated all reduce. -- 055c409 by Yunlong Liu <yunlongl@x.ai>: Creates a new e2e test target for GPU. -- 93934eb by Yunlong Liu <yunlongl@x.ai>: E2E test passed. Merging this change closes #18254 FUTURE_COPYBARA_INTEGRATE_REVIEW=#18254 from yliu120:ar-simplify 93934eb PiperOrigin-RevId: 685835366
copybara-service
bot
force-pushed
the
test_685835366
branch
from
8d42868
to
221ac9b
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Outdated suggestions cannot be applied.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.
PR #18254: [XLA:GPU] Simplifies All-Reduce if they can be simplified or are degenerated.
Imported from GitHub PR #18254
This PR adds back the AllReduceSimplifier pass to the GPU compiler as in some cases we see JAX programs will add some degenerated psum to the HLOs,
Code pointer in JAX:
The compiler should generally remove these unnecessary ARs so that it unblocks many optimizations like collective pipeliner. It also emits unncessary runtime thunks as well.
Copybara import of the project:
--
b89df55 by Yunlong Liu yunlongl@x.ai:
Removes degenerated all reduce.
--
055c409 by Yunlong Liu yunlongl@x.ai:
Creates a new e2e test target for GPU.
--
93934eb by Yunlong Liu yunlongl@x.ai:
E2E test passed.
Merging this change closes #18254
FUTURE_COPYBARA_INTEGRATE_REVIEW=#18254 from yliu120:ar-simplify 93934eb