[mxfp8 training] make alignment size configurable in benchmark script#4246
Merged
danielvegamyhre merged 1 commit intomainfrom Apr 7, 2026
Merged
[mxfp8 training] make alignment size configurable in benchmark script#4246danielvegamyhre merged 1 commit intomainfrom
danielvegamyhre merged 1 commit intomainfrom
Conversation
stack-info: PR: #4246, branch: danielvegamyhre/stack/163
danielvegamyhre
force-pushed
the
danielvegamyhre/stack/163
branch
from
2728f01 to
cfc5db9
Compare
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/4246
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ⏳ No Failures, 1 PendingAs of commit cfc5db9 with merge base d26bbae ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
danielvegamyhre
commented
Apr 6, 2026
| ] | ||
| recipes = [ | ||
| MXFP8TrainingRecipe.MXFP8_RCEIL, | ||
| MXFP8TrainingRecipe.MXFP8_RCEIL_WGRAD_WITH_HP, |
Contributor
Author
There was a problem hiding this comment.
I also am removing this from the default config to speed up the benchmark runs when iterating, since this is least used recipe / not the top priority. If I (or anyone) wants to check benchmarks for it they can easily add it to the config with one line of code.
jerryzh168
approved these changes
Apr 6, 2026
brucechanglongxu
pushed a commit
to brucechanglongxu/ao
that referenced
this pull request
Apr 9, 2026
…pytorch#4246) stack-info: PR: pytorch#4246, branch: danielvegamyhre/stack/163
brucechanglongxu
pushed a commit
to brucechanglongxu/ao
that referenced
this pull request
Apr 9, 2026
…pytorch#4246) stack-info: PR: pytorch#4246, branch: danielvegamyhre/stack/163
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
2 participants
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.
We should make token group alignment size configurable for benchmarking.
Context: I am experimenting with performance trade-offs of padding token group sizes to 128 instead of 32 to avoid expensive dynamic per group padding when writing mxfp8 scale factors to blocked layout in the backward pass for
wgradcalculation.