Add BF16 routing support to CUDA MoE - #32659
Merged
Merged
Conversation
Contributor
There was a problem hiding this comment.
🟡 Changes recommended
Add end-to-end BF16 sparse-mixer test coverage, including the CUDA capability guard.
Get a fresh assessment by requesting another Copilot review.
Pull request overview
Adds BF16 routing support to the CUDA MoE sparse mixer and softmax top-k paths.
Changes:
- Detects BF16 inputs.
- Routes BF16 tensors to the appropriate CUDA kernels.
File summaries
| File | Description |
|---|---|
onnxruntime/contrib_ops/cuda/moe/moe.cc |
Adds BF16 routing dispatch. |
Review details
- Files reviewed: 1/1 changed files
- Comments generated: 1
- Review effort level: Lite (auto)
Note
Copilot is running an experiment and ran this review at Lite.
💡 Configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Akshay Sonawane (apsonawane)
approved these changes
Sep 16, 2026
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
This pull request adds support for the
bfloat16(BF16) data type in the Mixture of Experts (MoE) CUDA kernel implementation. The main changes ensure that both the sparse mixer and softmax top-k operations can now handle BF16 input tensors, improving compatibility and performance on hardware supporting BF16.Support for BF16 data type:
IsDataType<BFloat16>()in the MoE kernel (moe.cc).LaunchSparseMixerTop2with BF16 data when appropriate.LaunchSoftmaxTopKwith BF16 data when appropriate.Motivation and Context
BF16 routing support was missing in the CUDA MoE kernel.