adjusting the communication method in graph mode#1194
Merged
ganyi1996ppo merged 1 commit intovllm-project:mainfrom Jun 25, 2025
Merged
adjusting the communication method in graph mode#1194ganyi1996ppo merged 1 commit intovllm-project:mainfrom
ganyi1996ppo merged 1 commit intovllm-project:mainfrom
Conversation
sharonyunyun
force-pushed
the
main
branch
3 times, most recently
from
b1227b4 to
9294612
Compare
Collaborator
|
I have cancelled the e2e test for quick fix deepseek problem. Please recheck the CI later. Sorry about this. |
Contributor
|
Could you add performance test logs and benefit result details to the PR description？ |
sharonyunyun
force-pushed
the
main
branch
5 times, most recently
from
983b820 to
6f640c7
Compare
sharonyunyun
changed the title
sharonyunyun
force-pushed
the
main
branch
2 times, most recently
from
8e79e3d to
7c32c3f
Compare
sharonyunyun
force-pushed
the
main
branch
6 times, most recently
from
235d1dc to
981a6df
Compare
Contributor
|
This pull request has conflicts, please resolve those before we can evaluate the pull request. |
sharonyunyun
force-pushed
the
main
branch
2 times, most recently
from
d635d22 to
16d9dce
Compare
sharonyunyun
force-pushed
the
main
branch
3 times, most recently
from
d6d17d2 to
f3affb8
Compare
Contributor
|
This pull request has conflicts, please resolve those before we can evaluate the pull request. |
sharonyunyun
force-pushed
the
main
branch
2 times, most recently
from
3d09b8e to
0b315f8
Compare
Yikun
added
long-term-test
sharonyunyun
force-pushed
the
main
branch
4 times, most recently
from
5e67447 to
cc59238
Compare
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #1194 +/- ##
==========================================
- Coverage 27.39% 27.04% -0.36%
==========================================
Files 56 56
Lines 6191 6275 +84
==========================================
+ Hits 1696 1697 +1
- Misses 4495 4578 +83
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:
|
Contributor
|
This pull request has conflicts, please resolve those before we can evaluate the pull request. |
sharonyunyun
force-pushed
the
main
branch
2 times, most recently
from
45d2385 to
0169aed
Compare
Signed-off-by: sharonyunyun <zhangying134@huawei.com>
wangxiyuan
approved these changes
Jun 25, 2025
ganyi1996ppo
approved these changes
Jun 25, 2025
weijinqian0
pushed a commit
to weijinqian0/vllm-ascend
that referenced
this pull request
Jun 30, 2025
### What this PR does / why we need it? Communication performance optimization: replace allreduce with reduce_scatter+all_gather in MLA layer's TP group，to remove stridedsliced and all_gather in MOE layer. when tp > 1, It is enabled during the decode phase of the graph mode when enable_multistream_moe、MLA, use_v1, and MC2 are used. According to the end-to-end RL inference test results, this PR can bring 3% gain in the decode stage. **Before Improvement** Profiling kernel_details  Evaluation   **After Improvement** Profiling kernel_details  Evaluation   ### Does this PR introduce _any_ user-facing change? Users need to configure enable_multistream_moe=True ### How was this patch tested? Add e2e test cases to cover code logic Signed-off-by: sharonyunyun <zhangying134@huawei.com>
chopper0126
pushed a commit
to chopper0126/vllm-ascend
that referenced
this pull request
Oct 16, 2025
### What this PR does / why we need it? Communication performance optimization: replace allreduce with reduce_scatter+all_gather in MLA layer's TP group，to remove stridedsliced and all_gather in MOE layer. when tp > 1, It is enabled during the decode phase of the graph mode when enable_multistream_moe、MLA, use_v1, and MC2 are used. According to the end-to-end RL inference test results, this PR can bring 3% gain in the decode stage. **Before Improvement** Profiling kernel_details  Evaluation   **After Improvement** Profiling kernel_details  Evaluation   ### Does this PR introduce _any_ user-facing change? Users need to configure enable_multistream_moe=True ### How was this patch tested? Add e2e test cases to cover code logic Signed-off-by: sharonyunyun <zhangying134@huawei.com>
Angazenn
pushed a commit
to Angazenn/vllm-ascend
that referenced
this pull request
Oct 21, 2025
### What this PR does / why we need it? Communication performance optimization: replace allreduce with reduce_scatter+all_gather in MLA layer's TP group，to remove stridedsliced and all_gather in MOE layer. when tp > 1, It is enabled during the decode phase of the graph mode when enable_multistream_moe、MLA, use_v1, and MC2 are used. According to the end-to-end RL inference test results, this PR can bring 3% gain in the decode stage. **Before Improvement** Profiling kernel_details  Evaluation   **After Improvement** Profiling kernel_details  Evaluation   ### Does this PR introduce _any_ user-facing change? Users need to configure enable_multistream_moe=True ### How was this patch tested? Add e2e test cases to cover code logic Signed-off-by: sharonyunyun <zhangying134@huawei.com>
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
5 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.
What this PR does / why we need it?
Communication performance optimization: replace allreduce with reduce_scatter+all_gather in MLA layer's TP group，to remove stridedsliced and all_gather in MOE layer.
when tp > 1, It is enabled during the decode phase of the graph mode when enable_multistream_moe、MLA, use_v1, and MC2 are used.
According to the end-to-end RL inference test results, this PR can bring 3% gain in the decode stage.
Before Improvement
Profiling kernel_details
Evaluation
After Improvement
Profiling kernel_details
Evaluation
Does this PR introduce any user-facing change?
Users need to configure enable_multistream_moe=True
How was this patch tested?
Add e2e test cases to cover code logic