[Feat] Shared expert dp for deepseek and deepseek_mtp by zhaozx-cn · Pull Request #3495 · vllm-project/vllm-ascend

zhaozx-cn · 2025-10-16T03:08:19Z

What this PR does / why we need it?

shared expert dp for deepseek and deepseek_mtp, could be combined with sp to improve performance.

How was this patch tested?

vLLM version: v0.11.0rc3
vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0

Signed-off-by: zhaozx-cn <zhaozx2116@163.com> Co-authored-by: realliujiaxu <realliujiaxu@163.com>

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

github-actions · 2025-10-16T03:08:29Z

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

A PR should do only one thing, smaller PRs enable faster reviews.
Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

gemini-code-assist

Code Review

This pull request refactors the handling of shared expert data parallelism by unifying it under the sequence parallelism flag (sp_enabled). It introduces a new custom op maybe_chunk_residual to correctly handle tensor shapes in residual connections when sequence parallelism is active. The changes also decouple the shared expert DP feature from the torchair backend, which seems intentional. The refactoring is clean and the implementation of the new logic appears correct. I have not found any critical or high-severity issues in these changes.

weijinqian0 · 2025-10-16T03:18:29Z

+        try:
+            forward_context = get_forward_context()
+            sp_enabled = forward_context.sp_enabled
+            if sp_enabled and self.debug_layer_idx < self.layers:


debug_layer_idx should be named as mtp_layer_idx

weijinqian0 · 2025-10-16T03:29:28Z

        # We retain the env VLLM_ASCEND_ENABLE_FLASHCOMM here for backward compatibility.
-        or bool(int(os.getenv("VLLM_ASCEND_ENABLE_FLASHCOMM", '0'))))
+        or bool(int(os.getenv("VLLM_ASCEND_ENABLE_FLASHCOMM", '0'))) or
+        get_ascend_config().enable_shared_expert_dp)


it may comfused users. when we set enable_shared_expert_dp, it will lead to set sp=1. But these two settings' meanings are different.

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

MengqingCao · 2025-10-17T02:35:28Z

plz update pr message

) shared expert dp for deepseek and deepseek_mtp, could be combined with sp to improve performance. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: zhaozx-cn <zhaozx2116@163.com> Co-authored-by: realliujiaxu <realliujiaxu@163.com>

) shared expert dp for deepseek and deepseek_mtp, could be combined with sp to improve performance. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: zhaozx-cn <zhaozx2116@163.com> Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: MrZ20 <2609716663@qq.com>

) ### What this PR does / why we need it? shared expert dp for deepseek and deepseek_mtp, could be combined with sp to improve performance. ### How was this patch tested? - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: zhaozx-cn <zhaozx2116@163.com> Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: MrZ20 <2609716663@qq.com>

) shared expert dp for deepseek and deepseek_mtp, could be combined with sp to improve performance. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: zhaozx-cn <zhaozx2116@163.com> Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: MrZ20 <2609716663@qq.com>

…roject#3495)" This reverts commit bf87606.

…roject#3495)" This reverts commit bf87606. Signed-off-by: linfeng-yuan <1102311262@qq.com>

#3586) ### What this PR does / why we need it? This reverts commit bf87606. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? E2E vllm serving with `enable_shared_expert_dp: true` in eager mode as before. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: linfeng-yuan <1102311262@qq.com>

…roject#3495)" (vllm-project#3586) This reverts commit vllm-project@bf87606. No. E2E vllm serving with `enable_shared_expert_dp: true` in eager mode as before. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: linfeng-yuan <1102311262@qq.com>

) ### What this PR does / why we need it? shared expert dp for deepseek and deepseek_mtp, could be combined with sp to improve performance. ### How was this patch tested? - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: zhaozx-cn <zhaozx2116@163.com> Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: luolun <luolun1995@cmbchina.com>

…roject#3495)" (vllm-project#3586) ### What this PR does / why we need it? This reverts commit vllm-project@bf87606. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? E2E vllm serving with `enable_shared_expert_dp: true` in eager mode as before. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: linfeng-yuan <1102311262@qq.com> Signed-off-by: luolun <luolun1995@cmbchina.com>

) ### What this PR does / why we need it? shared expert dp for deepseek and deepseek_mtp, could be combined with sp to improve performance. ### How was this patch tested? - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: zhaozx-cn <zhaozx2116@163.com> Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: hwhaokun <haokun0405@163.com>

…roject#3495)" (vllm-project#3586) ### What this PR does / why we need it? This reverts commit vllm-project@bf87606. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? E2E vllm serving with `enable_shared_expert_dp: true` in eager mode as before. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: linfeng-yuan <1102311262@qq.com> Signed-off-by: hwhaokun <haokun0405@163.com>

) ### What this PR does / why we need it? shared expert dp for deepseek and deepseek_mtp, could be combined with sp to improve performance. ### How was this patch tested? - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: zhaozx-cn <zhaozx2116@163.com> Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: nsdie <yeyifan@huawei.com>

…roject#3495)" (vllm-project#3586) ### What this PR does / why we need it? This reverts commit vllm-project@bf87606. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? E2E vllm serving with `enable_shared_expert_dp: true` in eager mode as before. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: linfeng-yuan <1102311262@qq.com> Signed-off-by: nsdie <yeyifan@huawei.com>

) ### What this PR does / why we need it? shared expert dp for deepseek and deepseek_mtp, could be combined with sp to improve performance. ### How was this patch tested? - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: zhaozx-cn <zhaozx2116@163.com> Co-authored-by: realliujiaxu <realliujiaxu@163.com>

…roject#3495)" (vllm-project#3586) ### What this PR does / why we need it? This reverts commit vllm-project@bf87606. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? E2E vllm serving with `enable_shared_expert_dp: true` in eager mode as before. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: linfeng-yuan <1102311262@qq.com>

@zzzzwwjj

I'd like to nominate @zzzzwwjj @realliujiaxu @LCAIZJ to join vLLM Ascend committer team. @zzzzwwjj --- - Review Quality‌: He has completed 80+reviews since April. 2025, include #3232 (comment), #4822 (comment), #4768 (comment) high quality review. - Sustained Contributions 15+ Valuable bug fix and refactor is very good. https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Azzzzwwjj+is%3Aclosed+review%3Aapproved Continuous optimization of code architecture https://github.com/vllm-project/vllm-ascend/pulls?q=author%3Azzzzwwjj+is%3Amerged - Quality Contribution‌: #1229 #1979 #4359 #4878 - Community Involvement‌: He lead the #1147, to refactor AscendFusedMoE at the first time. He shared topics about large-scale distributed inference and reinforcement learning on vLLM-Ascend meetup on August 2nd. @realliujiaxu --- - Review Quality‌: He has completed about [40+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Arealliujiaxu+-author%3Arealliujiaxu+) since September, include #4868 (comment), #2275 (comment). - Sustained Contributions He has completed (17 commits)[https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Arealliujiaxu+is%3Amerged], continuously optimizing the performance of the MoE model. - Quality Contribution‌: Contributed the Flash Comm1 feature to the community, supporting both eager and aclgraph execution modes, while compatible with multiple MoE models including DeepSeek and GLM4.5. - #3334 - #3420 - #3015 co-author: - #3495 - #4868 - Community Involvement‌: 1. Completed two major refactors, enabling vllm-ascend to evolve more rapidly and robustly: [Linear module](#2867) and [rejection sampler](#4975) 2. [fixed 8 bugs](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Arealliujiaxu+is%3Amerged+bugfix+) in graph mode, spec decoding and async scheduling. @LCAIZJ --- - Review Quality‌: He's been the go-to reviewer for virtually all PD disaggregation and KV Pool related PRs, having completed [30+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3ALCAIZJ+is%3Aopen+-author%3ALCAIZJ+) since May 2025. Notable examples include [discussion_r2553887360](#4345 (comment)), [issuecomment-3540994801](#4161 (comment)), and [discussion_r2492593988](#3981 (comment)), all demonstrating thorough and insightful feedback. - Sustained and Quality Contributions: His contributions reflect a strong grasp of both ‌vLLM‌ and ‌vLLM Ascend‌ codebases, particularly in prefill-decode disaggregation and KV pool areas ([7 PRs merged](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3ALCAIZJ+is%3Amerged+)). Prefill-Decode Disaggregation: Delivered KV transfer functionality using Mooncake TransferEngine and enabled layerwise KV transfer #1568 #2602 KV Pool: Developed the foundational KV Pool infrastructure and migrated it to the latest ADXL stack #2913 #3350 - Quality Contribution‌: #1568 #2602 #2913 #3350 - Community Involvement‌: He actively responds to [community issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20commenter%3ALCAIZJ%20is%3Aopen%20-author%3ALCAIZJ), continuously monitors functionality and accuracy issues related to PD disaggregation and KV Pool, and proactively delivers [bug fixes](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3ALCAIZJ+is%3Amerged+bugfix). - vLLM version: v0.12.0 - vLLM main: vllm-project/vllm@ad32e3e Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>

@zzzzwwjj

…t#5152) I'd like to nominate @zzzzwwjj @realliujiaxu @LCAIZJ to join vLLM Ascend committer team. @zzzzwwjj --- - Review Quality‌: He has completed 80+reviews since April. 2025, include vllm-project#3232 (comment), vllm-project#4822 (comment), vllm-project#4768 (comment) high quality review. - Sustained Contributions 15+ Valuable bug fix and refactor is very good. https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Azzzzwwjj+is%3Aclosed+review%3Aapproved Continuous optimization of code architecture https://github.com/vllm-project/vllm-ascend/pulls?q=author%3Azzzzwwjj+is%3Amerged - Quality Contribution‌: vllm-project#1229 vllm-project#1979 vllm-project#4359 vllm-project#4878 - Community Involvement‌: He lead the vllm-project#1147, to refactor AscendFusedMoE at the first time. He shared topics about large-scale distributed inference and reinforcement learning on vLLM-Ascend meetup on August 2nd. @realliujiaxu --- - Review Quality‌: He has completed about [40+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Arealliujiaxu+-author%3Arealliujiaxu+) since September, include vllm-project#4868 (comment), vllm-project#2275 (comment). - Sustained Contributions He has completed (17 commits)[https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Arealliujiaxu+is%3Amerged], continuously optimizing the performance of the MoE model. - Quality Contribution‌: Contributed the Flash Comm1 feature to the community, supporting both eager and aclgraph execution modes, while compatible with multiple MoE models including DeepSeek and GLM4.5. - vllm-project#3334 - vllm-project#3420 - vllm-project#3015 co-author: - vllm-project#3495 - vllm-project#4868 - Community Involvement‌: 1. Completed two major refactors, enabling vllm-ascend to evolve more rapidly and robustly: [Linear module](vllm-project#2867) and [rejection sampler](vllm-project#4975) 2. [fixed 8 bugs](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Arealliujiaxu+is%3Amerged+bugfix+) in graph mode, spec decoding and async scheduling. @LCAIZJ --- - Review Quality‌: He's been the go-to reviewer for virtually all PD disaggregation and KV Pool related PRs, having completed [30+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3ALCAIZJ+is%3Aopen+-author%3ALCAIZJ+) since May 2025. Notable examples include [discussion_r2553887360](vllm-project#4345 (comment)), [issuecomment-3540994801](vllm-project#4161 (comment)), and [discussion_r2492593988](vllm-project#3981 (comment)), all demonstrating thorough and insightful feedback. - Sustained and Quality Contributions: His contributions reflect a strong grasp of both ‌vLLM‌ and ‌vLLM Ascend‌ codebases, particularly in prefill-decode disaggregation and KV pool areas ([7 PRs merged](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3ALCAIZJ+is%3Amerged+)). Prefill-Decode Disaggregation: Delivered KV transfer functionality using Mooncake TransferEngine and enabled layerwise KV transfer vllm-project#1568 vllm-project#2602 KV Pool: Developed the foundational KV Pool infrastructure and migrated it to the latest ADXL stack vllm-project#2913 vllm-project#3350 - Quality Contribution‌: vllm-project#1568 vllm-project#2602 vllm-project#2913 vllm-project#3350 - Community Involvement‌: He actively responds to [community issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20commenter%3ALCAIZJ%20is%3Aopen%20-author%3ALCAIZJ), continuously monitors functionality and accuracy issues related to PD disaggregation and KV Pool, and proactively delivers [bug fixes](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3ALCAIZJ+is%3Amerged+bugfix). - vLLM version: v0.12.0 - vLLM main: vllm-project/vllm@ad32e3e Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>

@zzzzwwjj

…t#5152) I'd like to nominate @zzzzwwjj @realliujiaxu @LCAIZJ to join vLLM Ascend committer team. @zzzzwwjj --- - Review Quality‌: He has completed 80+reviews since April. 2025, include vllm-project#3232 (comment), vllm-project#4822 (comment), vllm-project#4768 (comment) high quality review. - Sustained Contributions 15+ Valuable bug fix and refactor is very good. https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Azzzzwwjj+is%3Aclosed+review%3Aapproved Continuous optimization of code architecture https://github.com/vllm-project/vllm-ascend/pulls?q=author%3Azzzzwwjj+is%3Amerged - Quality Contribution‌: vllm-project#1229 vllm-project#1979 vllm-project#4359 vllm-project#4878 - Community Involvement‌: He lead the vllm-project#1147, to refactor AscendFusedMoE at the first time. He shared topics about large-scale distributed inference and reinforcement learning on vLLM-Ascend meetup on August 2nd. @realliujiaxu --- - Review Quality‌: He has completed about [40+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Arealliujiaxu+-author%3Arealliujiaxu+) since September, include vllm-project#4868 (comment), vllm-project#2275 (comment). - Sustained Contributions He has completed (17 commits)[https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Arealliujiaxu+is%3Amerged], continuously optimizing the performance of the MoE model. - Quality Contribution‌: Contributed the Flash Comm1 feature to the community, supporting both eager and aclgraph execution modes, while compatible with multiple MoE models including DeepSeek and GLM4.5. - vllm-project#3334 - vllm-project#3420 - vllm-project#3015 co-author: - vllm-project#3495 - vllm-project#4868 - Community Involvement‌: 1. Completed two major refactors, enabling vllm-ascend to evolve more rapidly and robustly: [Linear module](vllm-project#2867) and [rejection sampler](vllm-project#4975) 2. [fixed 8 bugs](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Arealliujiaxu+is%3Amerged+bugfix+) in graph mode, spec decoding and async scheduling. @LCAIZJ --- - Review Quality‌: He's been the go-to reviewer for virtually all PD disaggregation and KV Pool related PRs, having completed [30+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3ALCAIZJ+is%3Aopen+-author%3ALCAIZJ+) since May 2025. Notable examples include [discussion_r2553887360](vllm-project#4345 (comment)), [issuecomment-3540994801](vllm-project#4161 (comment)), and [discussion_r2492593988](vllm-project#3981 (comment)), all demonstrating thorough and insightful feedback. - Sustained and Quality Contributions: His contributions reflect a strong grasp of both ‌vLLM‌ and ‌vLLM Ascend‌ codebases, particularly in prefill-decode disaggregation and KV pool areas ([7 PRs merged](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3ALCAIZJ+is%3Amerged+)). Prefill-Decode Disaggregation: Delivered KV transfer functionality using Mooncake TransferEngine and enabled layerwise KV transfer vllm-project#1568 vllm-project#2602 KV Pool: Developed the foundational KV Pool infrastructure and migrated it to the latest ADXL stack vllm-project#2913 vllm-project#3350 - Quality Contribution‌: vllm-project#1568 vllm-project#2602 vllm-project#2913 vllm-project#3350 - Community Involvement‌: He actively responds to [community issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20commenter%3ALCAIZJ%20is%3Aopen%20-author%3ALCAIZJ), continuously monitors functionality and accuracy issues related to PD disaggregation and KV Pool, and proactively delivers [bug fixes](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3ALCAIZJ+is%3Amerged+bugfix). - vLLM version: v0.12.0 - vLLM main: vllm-project/vllm@ad32e3e Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com> Signed-off-by: zrj026 <zhangrunjiang026@gmail.com>

@zzzzwwjj

…t#5152) I'd like to nominate @zzzzwwjj @realliujiaxu @LCAIZJ to join vLLM Ascend committer team. @zzzzwwjj --- - Review Quality‌: He has completed 80+reviews since April. 2025, include vllm-project#3232 (comment), vllm-project#4822 (comment), vllm-project#4768 (comment) high quality review. - Sustained Contributions 15+ Valuable bug fix and refactor is very good. https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Azzzzwwjj+is%3Aclosed+review%3Aapproved Continuous optimization of code architecture https://github.com/vllm-project/vllm-ascend/pulls?q=author%3Azzzzwwjj+is%3Amerged - Quality Contribution‌: vllm-project#1229 vllm-project#1979 vllm-project#4359 vllm-project#4878 - Community Involvement‌: He lead the vllm-project#1147, to refactor AscendFusedMoE at the first time. He shared topics about large-scale distributed inference and reinforcement learning on vLLM-Ascend meetup on August 2nd. @realliujiaxu --- - Review Quality‌: He has completed about [40+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Arealliujiaxu+-author%3Arealliujiaxu+) since September, include vllm-project#4868 (comment), vllm-project#2275 (comment). - Sustained Contributions He has completed (17 commits)[https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Arealliujiaxu+is%3Amerged], continuously optimizing the performance of the MoE model. - Quality Contribution‌: Contributed the Flash Comm1 feature to the community, supporting both eager and aclgraph execution modes, while compatible with multiple MoE models including DeepSeek and GLM4.5. - vllm-project#3334 - vllm-project#3420 - vllm-project#3015 co-author: - vllm-project#3495 - vllm-project#4868 - Community Involvement‌: 1. Completed two major refactors, enabling vllm-ascend to evolve more rapidly and robustly: [Linear module](vllm-project#2867) and [rejection sampler](vllm-project#4975) 2. [fixed 8 bugs](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Arealliujiaxu+is%3Amerged+bugfix+) in graph mode, spec decoding and async scheduling. @LCAIZJ --- - Review Quality‌: He's been the go-to reviewer for virtually all PD disaggregation and KV Pool related PRs, having completed [30+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3ALCAIZJ+is%3Aopen+-author%3ALCAIZJ+) since May 2025. Notable examples include [discussion_r2553887360](vllm-project#4345 (comment)), [issuecomment-3540994801](vllm-project#4161 (comment)), and [discussion_r2492593988](vllm-project#3981 (comment)), all demonstrating thorough and insightful feedback. - Sustained and Quality Contributions: His contributions reflect a strong grasp of both ‌vLLM‌ and ‌vLLM Ascend‌ codebases, particularly in prefill-decode disaggregation and KV pool areas ([7 PRs merged](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3ALCAIZJ+is%3Amerged+)). Prefill-Decode Disaggregation: Delivered KV transfer functionality using Mooncake TransferEngine and enabled layerwise KV transfer vllm-project#1568 vllm-project#2602 KV Pool: Developed the foundational KV Pool infrastructure and migrated it to the latest ADXL stack vllm-project#2913 vllm-project#3350 - Quality Contribution‌: vllm-project#1568 vllm-project#2602 vllm-project#2913 vllm-project#3350 - Community Involvement‌: He actively responds to [community issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20commenter%3ALCAIZJ%20is%3Aopen%20-author%3ALCAIZJ), continuously monitors functionality and accuracy issues related to PD disaggregation and KV Pool, and proactively delivers [bug fixes](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3ALCAIZJ+is%3Amerged+bugfix). - vLLM version: v0.12.0 - vLLM main: vllm-project/vllm@ad32e3e Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com> Signed-off-by: zrj026 <zhangrunjiang026@gmail.com>

zhaozx-cn and others added 2 commits October 16, 2025 10:49

shared expert dp for ds and ds_mtp

c71a05d

Signed-off-by: zhaozx-cn <zhaozx2116@163.com> Co-authored-by: realliujiaxu <realliujiaxu@163.com>

shared expert dp for ds and ds_mtp

1702d4d

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

github-actions bot added module:ops module:core labels Oct 16, 2025

gemini-code-assist bot reviewed Oct 16, 2025

View reviewed changes

weijinqian0 reviewed Oct 16, 2025

View reviewed changes

zhaozx-cn added 6 commits October 16, 2025 14:07

shared expert dp for ds and ds_mtp

01d17fe

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut

6e7c03f

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut

e2c9a18

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut

ec4246e

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut

9b48609

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut

56c2e22

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

github-actions bot added the module:tests label Oct 16, 2025

zhaozx-cn added 9 commits October 16, 2025 21:22

fix ut

ecd13cf

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut

2a53f85

Co-authored-by: realliujiaxu <realliujiaxu@163.com> Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut and mtp accuracy

5d4837d

Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut and mtp accuracy

16157f4

Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut and mtp accuracy

4e9f81e

Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut and mtp accuracy

029f162

Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut and mtp accuracy

106d141

Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut and mtp accuracy

69f2ce2

Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

fix ut and mtp accuracy

db31495

Signed-off-by: zhaozx-cn <zhaozx2116@163.com>

weijinqian0 approved these changes Oct 16, 2025

View reviewed changes

MengqingCao added the ready read for review label Oct 17, 2025

MengqingCao added the ready-for-test start test by label for PR label Oct 17, 2025

zhaozx-cn changed the title ~~shared expert dp for deepseek and deepseek_mtp~~ [Feat] Shared expert dp for deepseek and deepseek_mtp Oct 17, 2025

weijinqian0 merged commit bf87606 into vllm-project:main Oct 17, 2025
44 checks passed

linfeng-yuan added a commit to linfeng-yuan/vllm-ascend that referenced this pull request Oct 21, 2025

Revert "[Feat] Shared expert dp for deepseek and deepseek_mtp (vllm-p…

463a42a

…roject#3495)" This reverts commit bf87606.

linfeng-yuan added a commit to linfeng-yuan/vllm-ascend that referenced this pull request Oct 21, 2025

Revert "[Feat] Shared expert dp for deepseek and deepseek_mtp (vllm-p…

d27f82a

…roject#3495)" This reverts commit bf87606. Signed-off-by: linfeng-yuan <1102311262@qq.com>

linfeng-yuan added a commit to linfeng-yuan/vllm-ascend that referenced this pull request Oct 21, 2025

Revert "[Feat] Shared expert dp for deepseek and deepseek_mtp (vllm-p…

45a67fe

…roject#3495)" This reverts commit bf87606. Signed-off-by: linfeng-yuan <1102311262@qq.com>

zhaozx-cn deleted the sed_1015 branch November 20, 2025 06:52

wangxiyuan mentioned this pull request Dec 18, 2025

Nominate new maintainers @zzzzwwjj @realliujiaxu @LCAIZJ #5152

Merged

realliujiaxu mentioned this pull request Dec 23, 2025

[Misc]: Is FlashComm actually only used for SP-related AllGather operations? #5089

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[Feat] Shared expert dp for deepseek and deepseek_mtp#3495

[Feat] Shared expert dp for deepseek and deepseek_mtp#3495
weijinqian0 merged 17 commits intovllm-project:mainfrom
zhaozx-cn:sed_1015

zhaozx-cn commented Oct 16, 2025 •

edited

Loading

Uh oh!

github-actions bot commented Oct 16, 2025

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

weijinqian0 Oct 16, 2025

Uh oh!

weijinqian0 Oct 16, 2025 •

edited

Loading

Uh oh!

MengqingCao commented Oct 17, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

zhaozx-cn commented Oct 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What this PR does / why we need it?

How was this patch tested?

Uh oh!

github-actions bot commented Oct 16, 2025

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

weijinqian0 Oct 16, 2025

Choose a reason for hiding this comment

Uh oh!

weijinqian0 Oct 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MengqingCao commented Oct 17, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

zhaozx-cn commented Oct 16, 2025 •

edited

Loading

weijinqian0 Oct 16, 2025 •

edited

Loading