[Perf] Reduce memory usage by splitting tokens in fused_experts #1729
+125
−18
Conversation
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
Signed-off-by: ApsarasX <apsarax@outlook.com>
Signed-off-by: ApsarasX <apsarax@outlook.com>
github-actions
bot
added
documentation
Codecov ReportAttention: Patch coverage is
❌ Your patch check has failed because the patch coverage (9.67%) is below the target coverage (100.00%). You can increase the patch coverage or adjust the target coverage. Additional details and impacted files@@ Coverage Diff @@
## main #1729 +/- ##
===========================================
+ Coverage 27.39% 54.45% +27.06%
===========================================
Files 56 80 +24
Lines 6191 9995 +3804
===========================================
+ Hits 1696 5443 +3747
- Misses 4495 4552 +57
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
wangxiyuan
reviewed
Jul 11, 2025
| @@ -33,6 +33,7 @@ The following table lists the additional configuration options available in vLLM | |||
| | `expert_map_path` | str | `None` | When using expert load balancing for the MOE model, an expert map path needs to be passed in. | | |||
| | `chunked_prefill_for_mla` | bool | `False` | Whether to enable the fused operator-like chunked_prefill. | | |||
| | `kv_cache_dtype` | str | `None` | When using the kv cache quantization method, kv cache dtype needs to be set, currently only int8 is supported. | | |||
| | `fused_moe_max_chunk_size` | int | `max_num_batched_tokens * data_parallel_size` | The maximum token chunk size for the fused MoE operation. Input exceeding this size is split into multiple chunks for processing. | | |||
There was a problem hiding this comment.
Need test different case of data_parallel_size to make sure this change works as expect
|
DeepSeek-R1-W8A8, max_model_len=max_num_batched_tokens=32768
DeepSeek-R1-W8A8, max_model_len=32768, max_num_batched_tokens=8192
|
jianzs
approved these changes
Jul 17, 2025
|
This pull request has conflicts, please resolve those before we can evaluate the pull request. |
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.
What this PR does / why we need it?
Reduce activation memory usage during the prefill phase in MOE for long-context scenarios.
Does this PR introduce any user-facing change?
Yes. If the user wants to use this feature, they need to manually set the fused_moe_max_chunk_size field in the additional-config dictionary.
How was this patch tested?
Yes.