fix: drop orphaned sharded_moe - #31139
Closed
GaetanLepage wants to merge 1 commit into
Closed
Conversation
|
Azure Pipelines: There may be pipelines that require an authorized user to comment /azp run to run. |
13 tasks
Contributor
|
See another PR #31191 for this issue. |
Author
|
Indeed, sorry for the noise. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
Drop the orphaned ShardedMoE CUDA contrib op.
Motivation and Context
ShardedMoE's backend,
contrib_ops/cuda/moe/ft_moe/moe_kernel.hand theort_fastertransformer::CutlassMoeFCRunnerit wraps, was removed in the MoE GEMM refactor (#28467) which moved everything tocontrib_ops/cuda/llm/moe_gemm/.moe.cc/moe_base.hwere ported to the newonnxruntime::llm::kernels::cutlass_kernelsAPI, butsharded_moe.{h,cc}were not, so the file still includes the deleted header and fails to compile wheneveronnxruntime_USE_NCCLis enabled:The op has been dead (uncompilable) since that refactor.
Remove it along with its build/registration entries so NCCL builds again; all other NCCL collective ops (
AllReduce,AllGather,AllToAll,DistributedMatMul, ...) are unaffected.Related:
- #28467 (comment) (cc @SuperSandro2000)
- NixOS/nixpkgs#546053