Add QuantizedDBuffer for MXFP8 plane storage - #7265
Open
wujingyue wants to merge 2 commits into
Open
Conversation
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
Contributor
Author
|
/ok to test 654e29f |
Contributor
Author
|
/ok to test 654e29f |
Signed-off-by: Jingyue Wu <jingyuew@nvidia.com>
wujingyue
force-pushed
the
agent/mfsdp-quantized-dbuffer
branch
from
September 11, 2026 23:18
654e29f to
c1f8617
Compare
copy-pr-bot
Bot
force-pushed
the
pull-request/7257
branch
from
September 11, 2026 23:18
d270dc2 to
be18635
Compare
Contributor
Author
|
/ok to test c1f8617 |
Signed-off-by: Jingyue Wu <jingyuew@nvidia.com>
Contributor
Author
|
/ok to test ce7702f |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Split the buffer abstraction and its tests out of #7114. Depends on #7257 (DBuffer layout allocation), itself based on #6978.
Adds QuantizedDBuffer and its public export. It composes MXFP8 data and scale planes, derives scale ownership from the data layout, constructs TE wrappers, and supports quantization and redistribution. Includes buffer collective and scale-layout tests.
MFSDP parameter integration and end-to-end training tests remain in #7114, now stacked on this PR.
Validation: standalone test_quantized_dbuffer.py passed on two GB200 ranks (3 tests per rank).