WIP: Add PPU-native DeepGEMM BF16 unquantized MoE - #322
Draft
li199959 wants to merge 1 commit into
Draft
Conversation
Route unquantized BF16 MoE experts through deep_gemm's nopad grouped GEMM (m_grouped_gemm_bf16_bf16_bf16_nt_nopad + deepgemm_moe_permute, small-M decode auto-dispatches to the GEMV kernel) instead of FlagGems Triton fused_moe, on out-of-tree (PPU/thead) platforms. Opt-in via VLLM_FL_MOE=deepgemm; unset = unchanged. Also fixes an UnboundLocalError in _get_priority_backends for out-of-tree platforms when FlagGems is disabled. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
ppu-dev seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account. You have signed the CLA already but the status is still pending? Let us recheck it. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Add an opt-in PPU-native DeepGEMM path for unquantized BF16 MoE experts, enabled with
VLLM_FL_MOE=deepgemm.This draft PR is opened early to reserve the integration point and collect review feedback.
Current changes
DeepGemmExpertsFLusing nopad grouped GEMMBefore ready for review
maindeep_gemmValidation so far
main: cleangit diff --check: passedRuntime validation is still pending because the local inspection environment does not contain
torch,vllm, ordeep_gemm.