-
Notifications
You must be signed in to change notification settings - Fork 173
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Make MegaBlocks go vroom on Hopper. #24
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
dMoE benchmarks on 8x H100 with 8-way EMP:
|
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Outdated suggestions cannot be applied.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.
Add grouped GEMM-based dMoE to work around Triton limitations on SM90. Guard
turbo
use to we do not need it installed if quantization is not enabled. Add layer-wise dMoE benchmarks.After this PR, we recommend enabling
grouped_mlp
for SM90.grouped_mlp
should be used only with expert model parallelism to keep per-device expert counts low, which is important for efficiency with the current cuBLAS-based grouped GEMM kernels.