add permute fusion into hybrid ep#4089
Conversation
|
This PR has been automatically converted to draft because all PRs must start as drafts. When you are ready for review, click Ready for Review to begin the review process. This will:
See the contribution guide for more details. |
| moe_hybridep_num_sms: int = 16 | ||
| """Number of SMs to use for HybridEP. In pure NVL scenarios, | ||
| 16 SMs can generally achieve good bandwidth.""" | ||
| moe_hybridep_num_sms: Optional[int] = None |
There was a problem hiding this comment.
nit; could we add the defaults used in the doc string here.
There was a problem hiding this comment.
I'm not tend to do this, as the default number of SMs used on the hybrid EP side may change.
|
/ok to test 1f0733c |
|
/ok to test e28bb0e |
|
/ok to test e28bb0e |
|
/ok to test d31d205 |
|
/ok to test d31d205 |
|
/ok to test d31d205 |
1 similar comment
|
/ok to test d31d205 |
|
/ok to test 022d94b |
|
/ok to test 022d94b |
1 similar comment
|
/ok to test 022d94b |
|
/ok to test 139778d |
|
🔄 Merge queue validation started! You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/25027627614 |
Co-authored-by: root <root@eos0047.eos.clusters.nvidia.com> Co-authored-by: root <root@eos0260.eos.clusters.nvidia.com> Co-authored-by: Dennis(Zhenhuan) Liu <denliu@nvidia.com>
What does this PR do ?
This PR introduce the new feature: fuse the permute/unpermute with the dispatch/combine into 1 kernel
This feature is provided by the hybrid-ep
related PR om dev: #4073
Contribution process
Pre-checks
Code review
Feel free to message or comment the @mcore-oncall to help accelerate your merge into main. The less complex your PR is, the faster it will be approved and merged!
All PRs start as draft. If you open a non-draft PR, it will be automatically converted to draft.
Step 1: Mark PR as "Ready for Review"
.github/CODEOWNERS.Final Review might get declined if these requirements are not fulfilled.
Step 2: Final Review
For PRs that change
megatron/core, once all expert reviewers have approved, theFinal Reviewlabel is applied automatically and final reviewers are assigned.For PRs outside
megatron/core, this step is skipped.Step 3: Approved
Once all required reviewers have approved, the
Approvedlabel is applied automatically.Merge
Any member of mcore-engineers will be able to merge your PR.
For MRs into `dev` branch
The proposed review process for `dev` branch is under active discussion.MRs are mergable after one approval by either
eharper@nvidia.comorzijiey@nvidia.com.