Fix ragged all-to-all with ragged buffer factor in DeepSeek-V3.#4397
Open
copybara-service[bot] wants to merge 1 commit into
Open
Fix ragged all-to-all with ragged buffer factor in DeepSeek-V3.#4397copybara-service[bot] wants to merge 1 commit into
copybara-service[bot] wants to merge 1 commit into
Conversation
copybara-service
Bot
requested review from
RissyRan,
darisoy,
gagika,
michelle-yooh,
richjames0,
shralex and
suexu1025
as code owners
copybara-service
Bot
force-pushed
the
test_944752214
branch
from
03bef94 to
7049d6b
Compare
- Implement traffic matrix truncation in forward/backward all-to-all passes when `ragged_buffer_factor > 0` to prevent out-of-bounds writes in `ragged_all_to_all`. - Partition GMM v2 VMEM scratchpad (`partial_out_ref`) by N program ID to prevent conflicts when `num_n > 1`. - Add unit tests for the truncation logic. PiperOrigin-RevId: 944752214
copybara-service
Bot
force-pushed
the
test_944752214
branch
from
7049d6b to
d7c3a0f
Compare
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fix ragged all-to-all with ragged buffer factor in DeepSeek-V3.
ragged_buffer_factor > 0to prevent out-of-bounds writes inragged_all_to_all.partial_out_ref) by N program ID to prevent conflicts whennum_n > 1.