[Main] [Refactor] Enable MoECommMethod in Eager Mode #2791

Pr0Wh1teGivee · 2025-09-06T02:56:35Z

What this PR does / why we need it?

Replace prepare/finalize operation in fused_moe.py by moe_comm_method.prepare()/finalize()
Replace unified_fused_experts by moe_comm_method.fused_experts() in fused_moe.py/w8a8_dynamic.py/w4a8_dynamic.py
Add calling _select_moe_comm_method in spec-decode proposers.
Currently, w4a8_dynamic does not support gatherep, use all2allv instead.
Remove redundant code.

Does this PR introduce any user-facing change?

AllgatherEP switch is disabled in aclgraph/eager mode, just follow the rules in modelrunner_v1._select_moe_comm_method()

How was this patch tested?

e2e & ut

vLLM version: v0.10.2
vLLM main: vllm-project/vllm@7f6f2c1

github-actions · 2025-09-06T02:56:42Z

👋 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 Fused MoE implementation by restructuring the code into more modular and object-oriented components. The changes are extensive, moving logic from vllm_ascend/distributed/moe_comm_method.py and vllm_ascend/ops/common_fused_moe.py into new classes like MoECommMethod, FusedMoEPrepareAndFinalize, and various TokenDispatcher implementations under a new vllm_ascend/ops/moe/ directory. This improves code organization and separation of concerns. New unit tests have been added for the refactored components.

My main concern is with the NativeAllGatherCommImpl, which appears to be broken by the refactoring, containing dead code and not providing the native PyTorch fallback it's intended to. This should be addressed to avoid confusion and future bugs.

gemini-code-assist · 2025-09-06T02:58:25Z

vllm_ascend/ops/moe/moe_comm_method.py

+class NativeAllGatherCommImpl(AllGatherCommImpl):
+    """This implementation should be compatible with all scenarios.
+
+    Note that this implementation purely consists of native PyTorch ops
+    and does not use any NPU-specific ops. So the performance may not be optimal.
+    But it is a good fallback for scenarios where NPU-specific ops are not available.
+    """
+
+    def permute(
+        self,
+        hidden_states: torch.Tensor,
+        topk_ids: torch.Tensor,
+        topk_weights: torch.Tensor,
+        expert_map: torch.Tensor,
+        num_experts: int,
+        apply_a8_quantization: bool,
+    ) -> tuple[torch.Tensor, torch.Tensor, Optional[torch.Tensor], int]:
+        num_tokens = hidden_states.shape[0]
+
+        # Generate token indices and flatten
+        token_indices = torch.arange(num_tokens,
+                                     device=hidden_states.device,
+                                     dtype=torch.int64)
+        token_indices = (token_indices.unsqueeze(1).expand(
+            -1, self.moe_config.experts_per_token).reshape(-1))
+
+        # Flatten token-to-expert mappings and map to local experts
+        weights_flat = topk_weights.view(-1)
+        experts_flat = topk_ids.view(-1)
+        local_experts_flat = (expert_map[experts_flat]
+                              if expert_map is not None else experts_flat)
+
+        # Filter valid token-expert pairs
+        mask = local_experts_flat != -1
+        # FIXME: npu_grouped_matmul output random values at [num_valid_tokens:, ...]
+        # So we need to filter out invalid tokens by zeroing their weights.
+        # This is a workaround and should be removed after the issue is fixed
+        filtered_weights = torch.where(mask, weights_flat,
+                                       torch.zeros_like(weights_flat)).to(
+                                           topk_weights.dtype)
+        filtered_experts = torch.where(
+            mask,
+            local_experts_flat,
+            torch.full_like(local_experts_flat, num_experts),
+        ).to(topk_ids.dtype)
+
+        # Sort by local expert IDs
+        sort_indices = torch.argsort(filtered_experts.view(torch.float32))
+        self.sorted_token_indices = token_indices[sort_indices]
+        self.sorted_weights = filtered_weights[sort_indices]
+
+        # Compute token counts with minlength of num_experts
+        # This is equivalent to but faster than:
+        # >>> token_counts = torch.bincount(filtered_experts, minlength=num_experts)[:-1]
+        token_counts = torch.zeros(num_experts + 1,
+                                   device=hidden_states.device,
+                                   dtype=torch.int64)
+        ones = torch.ones_like(filtered_experts, dtype=torch.int64)
+        token_counts.scatter_add_(0, filtered_experts.to(torch.int64), ones)
+        expert_tokens = token_counts[:num_experts]
+
+        # Rearrange hidden_states
+        permuted_hidden_states = hidden_states[self.sorted_token_indices]
+
+        group_list_type = 1  # `count` mode
+
+        return permuted_hidden_states, expert_tokens, None, group_list_type
+
+    def unpermute(self, mlp_output: torch.Tensor,
+                  hidden_states: torch.Tensor) -> None:
+        mlp_output = mlp_output * self.sorted_weights.unsqueeze(1)
+
+        final_hidden_states = torch.zeros_like(hidden_states)
+        final_hidden_states.index_add_(0, self.sorted_token_indices,
+                                       mlp_output)
+
+        hidden_states[:] = final_hidden_states
+


The permute and unpermute methods within NativeAllGatherCommImpl appear to be dead code after the refactoring. The fused_experts method from the base MoECommMethod is now used, which calls a token_dispatcher instead of these methods.

Furthermore, NativeAllGatherCommImpl inherits its _get_token_dispatcher method from AllGatherCommImpl, which provides TokenDispatcherWithAllGather. This dispatcher uses NPU-specific operations, contradicting the docstring of NativeAllGatherCommImpl which states it should only use native PyTorch ops. This makes the class non-functional as a native fallback.

To fix this, these unused methods should be removed. A proper native fallback would require a new NativeTokenDispatcher that implements the logic from the old permute and unpermute methods, and NativeAllGatherCommImpl should be updated to use it.

github-actions · 2025-09-08T01:25:10Z

This pull request has conflicts, please resolve those before we can evaluate the pull request.

github-actions · 2025-09-09T12:40:49Z

This pull request has conflicts, please resolve those before we can evaluate the pull request.

github-actions · 2025-09-11T03:44:08Z

This pull request has conflicts, please resolve those before we can evaluate the pull request.

codecov · 2025-09-12T03:18:20Z

Codecov Report

❌ Patch coverage is 85.60000% with 18 lines in your changes missing coverage. Please review.
✅ Project coverage is 74.75%. Comparing base (1bbb20e) to head (1df0ea7).
⚠️ Report is 31 commits behind head on main.

Files with missing lines	Patch %	Lines
...m_ascend/ops/moe/fused_moe_prepare_and_finalize.py	88.88%	4 Missing ⚠️
vllm_ascend/worker/model_runner_v1.py	40.00%	3 Missing ⚠️
tests/ut/ops/test_fused_ops.py	89.47%	2 Missing ⚠️
vllm_ascend/ops/moe/token_dispatcher.py	33.33%	2 Missing ⚠️
vllm_ascend/quantization/w4a8_dynamic.py	0.00%	2 Missing ⚠️
vllm_ascend/quantization/w8a8_dynamic.py	0.00%	2 Missing ⚠️
vllm_ascend/spec_decode/mtp_proposer.py	0.00%	2 Missing ⚠️
vllm_ascend/ops/common_fused_moe.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2791      +/-   ##
==========================================
- Coverage   74.76%   74.75%   -0.02%     
==========================================
  Files         150      153       +3     
  Lines       20891    21080     +189     
==========================================
+ Hits        15620    15758     +138     
- Misses       5271     5322      +51

Flag	Coverage Δ
unittests	`74.75% <85.60%> (-0.02%)`	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

weijinqian0 · 2025-09-15T01:06:37Z

approved

Co-Authored-By: weijinqian0 <[email protected]> Signed-off-by: Pr0Wh1teGivee <[email protected]>

wangxiyuan · 2025-09-16T01:20:30Z

vllm_ascend/ops/moe/moe_mlp.py

    _output_dtype = w2_scale.dtype

-    is_mc2 = get_forward_context().fused_moe_state == FusedMoEState.MC2
+    is_mc2 = get_forward_context().moe_comm_method_name == "mc2commimpl"


there are a lot hardcode xxxcommimpl, it's good to use Enum instead.

Fair point, will be refactored in my next pr.

wangxiyuan · 2025-09-16T01:56:31Z

vllm_ascend/ops/fused_moe.py

-                    router_logits = self.naive_multicast(
-                        router_logits, cu_tokens_across_dp_cpu)
+        moe_comm_method_name = forward_context.moe_comm_method_name
+        forward_context.moe_comm_method = getattr(self, moe_comm_method_name)


set a new property to forward_context here is not suitable, it breaks the princeple of forward_context and make the code hard for debug. While i notice that it's been done in common fused moe already. Let's refactor it in the next PR.

Fair point, will be refactored in my next pr.

wangxiyuan · 2025-09-16T02:05:21Z

vllm_ascend/worker/model_runner_v1.py

        elif soc_version in {AscendSocVersion.A3}:
            moe_comm_method = "mc2" if num_tokens <= self.mc2_tokens_capacity else "alltoall"
        else:
            raise ValueError(f"Unsupported soc_version: {soc_version}")


looks that 310p is missing

FusedExpertsWithMoGE will be used by judging "model_type == "PanguProMoE" in the future. No need to add 310P here.

wangxiyuan · 2025-09-16T02:18:30Z

vllm_ascend/worker/model_runner_v1.py

        if not self.parallel_config.enable_expert_parallel:
            moe_comm_method = "allgather"
        elif soc_version in {AscendSocVersion.A2}:
            if num_tokens <= self.mc2_tokens_capacity and self.parallel_config.world_size_across_dp >= 16:


previous: go into allgather if VLLM_ENABLE_FUSED_EXPERTS_ALLGATHER_EP enabled
now: go into allgather if num_tokens <= self.mc2_tokens_capacity and self.parallel_config.world_size_across_dp >= 16 is False.

is the case squal?

Confirmed with Ant, it is ok to use _select_comm_method.

wangxiyuan

All the review comment can be fixed in the next PR

wangxiyuan · 2025-09-16T03:06:22Z

pangu is broken by qwen3-next PR. issue is here: #2949 We'll fix it later.

### What this PR does / why we need it? Fix issues mentioned in #2791 and some minor refactoring. 1. Use Enum instead of string. 2. Avoid setting a new property to forward_context in AscendFusedMoE.forward(). 3. Enabling TokenDispatcherWithMoge. 4. Remove redundant code. ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? Qwen3-30B-A3B/Qwen3-30B-A3B-W8A8/DeepSeek-V3-W4A8-Pruing/deepseek-mtp/pangu-pro-moe-pruing: 1. Enable/Disable EP 2. Aclgraph & eager - vLLM version: v0.10.2 - vLLM main: vllm-project/vllm@9607d5e Signed-off-by: Pr0Wh1teGivee <[email protected]> Co-authored-by: weijinqian0 <[email protected]>

…ject#3001) ### What this PR does / why we need it? Fix issues mentioned in vllm-project#2791 and some minor refactoring. 1. Use Enum instead of string. 2. Avoid setting a new property to forward_context in AscendFusedMoE.forward(). 3. Enabling TokenDispatcherWithMoge. 4. Remove redundant code. ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? Qwen3-30B-A3B/Qwen3-30B-A3B-W8A8/DeepSeek-V3-W4A8-Pruing/deepseek-mtp/pangu-pro-moe-pruing: 1. Enable/Disable EP 2. Aclgraph & eager - vLLM version: v0.10.2 - vLLM main: vllm-project/vllm@9607d5e Signed-off-by: Pr0Wh1teGivee <[email protected]> Co-authored-by: weijinqian0 <[email protected]> Signed-off-by: Che Ruan <[email protected]>

### What this PR does / why we need it? 1. Replace prepare/finalize operation in fused_moe.py by moe_comm_method.prepare()/finalize() 2. Replace unified_fused_experts by moe_comm_method.fused_experts() in fused_moe.py/w8a8_dynamic.py/w4a8_dynamic.py 3. Add calling _select_moe_comm_method in spec-decode proposers. 4. Currently, w4a8_dynamic does not support gatherep, use all2allv instead. 5. Remove redundant code. ### Does this PR introduce _any_ user-facing change? AllgatherEP switch is disabled in aclgraph/eager mode, just follow the rules in modelrunner_v1._select_moe_comm_method() ### How was this patch tested? e2e & ut - vLLM version: v0.10.2 - vLLM main: vllm-project/vllm@7f6f2c1 Signed-off-by: Pr0Wh1teGivee <[email protected]> Co-authored-by: weijinqian0 <[email protected]>

…nalinaly

…nalinaly (#3406) I'd like to nominate 4 new maintainers for vllm-ascend: ---- Yizhou Liu [@yiz-liu](https://github.com/yiz-liu) ---- **Review Quality‌**: He has completed [40+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Ayiz-liu) and provided solutions or guides for [10+ issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20commenter%3Ayiz-liu), which includes many quality review like [#issue-3428408401](#3002 (comment)), [#discussion_r2224572309](#1803 (comment)), [#issuecomment-2982470226](#1261 (comment)), [#issuecomment-2903621197](#836 (comment)), [#issuecomment-2857678691](#778 (comment)). **Sustained and High-Quality Contributions:** He has contributed more than [30+ commits](https://github.com/vllm-project/vllm-ascend/commits?author=yiz-liu) since Mar.2025, especially, aclgraph, DP, and EP related contributions are the main reason why I nominated him. As the owner of aclgraph support, he continuously improves aclgraph stability and performance as well as fixes key bugs. he laid the groundwork for EP-related functionality and delivered multiple foundational improvements **Community involvement:** He has a very good habit of logging issues：#1649 and is also very active and involved in [many issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20state%3Aopen%20commenter%3Ayiz-liu%20-author%3Ayiz-liu) to help users resolve issues. ---- Peng Yu [@paulyu12](https://github.com/paulyu12) --- The main reasons for his nomination are his expertise and key contributions to the LORA and sustained and major contributions (initial support/doc/bugfix) around Lora. **Sustained and Major Contributions:** @paulyu12 starts his contribution with [Lora and Mulit-Lora support](697908f) since Apr 2025, he contributed about [10+ commits and bugfixes](697908f) on vllm-ascend. **Review Quality‌ and Community Involvement‌:** He also helped more than 10+ users address [Lora related issues](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Apaulyu12+-author%3Apaulyu12+is%3Aclosed). I believe his addition will further improve vLLM Ascend Lora support. ---- Jinqian Wei [@weijinqian0](https://github.com/weijinqian0) --- The main reasons for his nomination are his key contributions to the RL scene and the high quality of his code reviews. **Review Quality‌:** He has completed [60+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Aweijinqian0+is%3Aopen+-author%3Aweijinqian0) since June. 2025, include [#comment-3284055430](#2791 (comment)), [discussion_r2332166704](#2817 (comment)), [discussion_r2343289692](#2846 (comment)) high quality review. **Sustained and Quality Contributions:** He has Deep understanding of ‌vLLM‌ and ‌vLLM Ascend‌ codebases and solid contributions in RL scene (about [10+ PR merged](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Aweijinqian0+is%3Amerged+) and 10+ PRs merged as co-author. - Code Refactor: As a co-author, he participated in the refactoring of the MOE module #2150 #2706 #2867 - Performance Enhancement for RL: Participated as a co-author in the design and development of the solution, contributing to the planning of core capabilities. #1547 #2120 and so on. So I think he's a great addition to the vLLM Ascend Maintainer team. ---- Chuanyu Qin [@nalinaly](https://github.com/nalinaly) --- The main reason I nominated Qinchuanyu is because he is the initial designer of aclgraph and torch-npu, two key components of vllm-ascend. Considering aclgraph will eventually become the main path for vllm-ascend's graph model, I propose to nominate him. **Sustained and Major Contributions:** In fact, chuanyu actively helped the users/developers of vllm-ascend since Mar 2025 ([vllm-discuss#162](https://discuss.vllm.ai/t/can-ascend-officially-draft-a-documentation-on-the-vllm-ascend-adaptation-for-graph-mode/162/5)), and also helped early users of vllm-ascend understand aclgraph. He provided lots of help in the process of integrating aclgraph with vllm-ascend. **Community Involvement‌:** As speaker, he also presents help users understand aclgraph and torch_npu [《The design philosophy of torch_npu and the high performance principle of aclGraph》](https://github.com/PyTorch-China/pytorch-meetup/blob/main/beijing-2025/%E3%80%905%E3%80%91torch_npu%20%E7%9A%84%E8%AE%BE%E8%AE%A1%E5%93%B2%E5%AD%A6%E4%B8%8E%20aclGraph%20%E9%AB%98%E6%80%A7%E8%83%BD%E5%8E%9F%E7%90%86-%E7%A7%A6%E4%BC%A0%E7%91%9C-0920.pdf) ---- They have activate contribution to vllm-ascend or have rich experience for ascend AI. Welcome! - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: wangxiyuan <[email protected]>

### What this PR does / why we need it? 1. Replace prepare/finalize operation in fused_moe.py by moe_comm_method.prepare()/finalize() 2. Replace unified_fused_experts by moe_comm_method.fused_experts() in fused_moe.py/w8a8_dynamic.py/w4a8_dynamic.py 3. Add calling _select_moe_comm_method in spec-decode proposers. 4. Currently, w4a8_dynamic does not support gatherep, use all2allv instead. 5. Remove redundant code. ### Does this PR introduce _any_ user-facing change? AllgatherEP switch is disabled in aclgraph/eager mode, just follow the rules in modelrunner_v1._select_moe_comm_method() ### How was this patch tested? e2e & ut - vLLM version: v0.10.2 - vLLM main: vllm-project/vllm@7f6f2c1 Signed-off-by: Pr0Wh1teGivee <[email protected]> Co-authored-by: weijinqian0 <[email protected]>

…ject#3001) ### What this PR does / why we need it? Fix issues mentioned in vllm-project#2791 and some minor refactoring. 1. Use Enum instead of string. 2. Avoid setting a new property to forward_context in AscendFusedMoE.forward(). 3. Enabling TokenDispatcherWithMoge. 4. Remove redundant code. ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? Qwen3-30B-A3B/Qwen3-30B-A3B-W8A8/DeepSeek-V3-W4A8-Pruing/deepseek-mtp/pangu-pro-moe-pruing: 1. Enable/Disable EP 2. Aclgraph & eager - vLLM version: v0.10.2 - vLLM main: vllm-project/vllm@9607d5e Signed-off-by: Pr0Wh1teGivee <[email protected]> Co-authored-by: weijinqian0 <[email protected]>

…nalinaly

…nalinaly (vllm-project#3406) I'd like to nominate 4 new maintainers for vllm-ascend: ---- Yizhou Liu [@yiz-liu](https://github.com/yiz-liu) ---- **Review Quality‌**: He has completed [40+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Ayiz-liu) and provided solutions or guides for [10+ issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20commenter%3Ayiz-liu), which includes many quality review like [#issue-3428408401](vllm-project#3002 (comment)), [#discussion_r2224572309](vllm-project#1803 (comment)), [#issuecomment-2982470226](vllm-project#1261 (comment)), [#issuecomment-2903621197](vllm-project#836 (comment)), [#issuecomment-2857678691](vllm-project#778 (comment)). **Sustained and High-Quality Contributions:** He has contributed more than [30+ commits](https://github.com/vllm-project/vllm-ascend/commits?author=yiz-liu) since Mar.2025, especially, aclgraph, DP, and EP related contributions are the main reason why I nominated him. As the owner of aclgraph support, he continuously improves aclgraph stability and performance as well as fixes key bugs. he laid the groundwork for EP-related functionality and delivered multiple foundational improvements **Community involvement:** He has a very good habit of logging issues：vllm-project#1649 and is also very active and involved in [many issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20state%3Aopen%20commenter%3Ayiz-liu%20-author%3Ayiz-liu) to help users resolve issues. ---- Peng Yu [@paulyu12](https://github.com/paulyu12) --- The main reasons for his nomination are his expertise and key contributions to the LORA and sustained and major contributions (initial support/doc/bugfix) around Lora. **Sustained and Major Contributions:** @paulyu12 starts his contribution with [Lora and Mulit-Lora support](vllm-project@697908f) since Apr 2025, he contributed about [10+ commits and bugfixes](vllm-project@697908f) on vllm-ascend. **Review Quality‌ and Community Involvement‌:** He also helped more than 10+ users address [Lora related issues](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Apaulyu12+-author%3Apaulyu12+is%3Aclosed). I believe his addition will further improve vLLM Ascend Lora support. ---- Jinqian Wei [@weijinqian0](https://github.com/weijinqian0) --- The main reasons for his nomination are his key contributions to the RL scene and the high quality of his code reviews. **Review Quality‌:** He has completed [60+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Aweijinqian0+is%3Aopen+-author%3Aweijinqian0) since June. 2025, include [#comment-3284055430](vllm-project#2791 (comment)), [discussion_r2332166704](vllm-project#2817 (comment)), [discussion_r2343289692](vllm-project#2846 (comment)) high quality review. **Sustained and Quality Contributions:** He has Deep understanding of ‌vLLM‌ and ‌vLLM Ascend‌ codebases and solid contributions in RL scene (about [10+ PR merged](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Aweijinqian0+is%3Amerged+) and 10+ PRs merged as co-author. - Code Refactor: As a co-author, he participated in the refactoring of the MOE module vllm-project#2150 vllm-project#2706 vllm-project#2867 - Performance Enhancement for RL: Participated as a co-author in the design and development of the solution, contributing to the planning of core capabilities. vllm-project#1547 vllm-project#2120 and so on. So I think he's a great addition to the vLLM Ascend Maintainer team. ---- Chuanyu Qin [@nalinaly](https://github.com/nalinaly) --- The main reason I nominated Qinchuanyu is because he is the initial designer of aclgraph and torch-npu, two key components of vllm-ascend. Considering aclgraph will eventually become the main path for vllm-ascend's graph model, I propose to nominate him. **Sustained and Major Contributions:** In fact, chuanyu actively helped the users/developers of vllm-ascend since Mar 2025 ([vllm-discuss#162](https://discuss.vllm.ai/t/can-ascend-officially-draft-a-documentation-on-the-vllm-ascend-adaptation-for-graph-mode/162/5)), and also helped early users of vllm-ascend understand aclgraph. He provided lots of help in the process of integrating aclgraph with vllm-ascend. **Community Involvement‌:** As speaker, he also presents help users understand aclgraph and torch_npu [《The design philosophy of torch_npu and the high performance principle of aclGraph》](https://github.com/PyTorch-China/pytorch-meetup/blob/main/beijing-2025/%E3%80%905%E3%80%91torch_npu%20%E7%9A%84%E8%AE%BE%E8%AE%A1%E5%93%B2%E5%AD%A6%E4%B8%8E%20aclGraph%20%E9%AB%98%E6%80%A7%E8%83%BD%E5%8E%9F%E7%90%86-%E7%A7%A6%E4%BC%A0%E7%91%9C-0920.pdf) ---- They have activate contribution to vllm-ascend or have rich experience for ascend AI. Welcome! - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: wangxiyuan <[email protected]>

…nalinaly

…nalinaly (vllm-project#3406) I'd like to nominate 4 new maintainers for vllm-ascend: ---- Yizhou Liu [@yiz-liu](https://github.com/yiz-liu) ---- **Review Quality‌**: He has completed [40+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Ayiz-liu) and provided solutions or guides for [10+ issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20commenter%3Ayiz-liu), which includes many quality review like [#issue-3428408401](vllm-project#3002 (comment)), [#discussion_r2224572309](vllm-project#1803 (comment)), [#issuecomment-2982470226](vllm-project#1261 (comment)), [#issuecomment-2903621197](vllm-project#836 (comment)), [#issuecomment-2857678691](vllm-project#778 (comment)). **Sustained and High-Quality Contributions:** He has contributed more than [30+ commits](https://github.com/vllm-project/vllm-ascend/commits?author=yiz-liu) since Mar.2025, especially, aclgraph, DP, and EP related contributions are the main reason why I nominated him. As the owner of aclgraph support, he continuously improves aclgraph stability and performance as well as fixes key bugs. he laid the groundwork for EP-related functionality and delivered multiple foundational improvements **Community involvement:** He has a very good habit of logging issues：vllm-project#1649 and is also very active and involved in [many issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20state%3Aopen%20commenter%3Ayiz-liu%20-author%3Ayiz-liu) to help users resolve issues. ---- Peng Yu [@paulyu12](https://github.com/paulyu12) --- The main reasons for his nomination are his expertise and key contributions to the LORA and sustained and major contributions (initial support/doc/bugfix) around Lora. **Sustained and Major Contributions:** @paulyu12 starts his contribution with [Lora and Mulit-Lora support](vllm-project@697908f) since Apr 2025, he contributed about [10+ commits and bugfixes](vllm-project@697908f) on vllm-ascend. **Review Quality‌ and Community Involvement‌:** He also helped more than 10+ users address [Lora related issues](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Apaulyu12+-author%3Apaulyu12+is%3Aclosed). I believe his addition will further improve vLLM Ascend Lora support. ---- Jinqian Wei [@weijinqian0](https://github.com/weijinqian0) --- The main reasons for his nomination are his key contributions to the RL scene and the high quality of his code reviews. **Review Quality‌:** He has completed [60+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Aweijinqian0+is%3Aopen+-author%3Aweijinqian0) since June. 2025, include [#comment-3284055430](vllm-project#2791 (comment)), [discussion_r2332166704](vllm-project#2817 (comment)), [discussion_r2343289692](vllm-project#2846 (comment)) high quality review. **Sustained and Quality Contributions:** He has Deep understanding of ‌vLLM‌ and ‌vLLM Ascend‌ codebases and solid contributions in RL scene (about [10+ PR merged](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Aweijinqian0+is%3Amerged+) and 10+ PRs merged as co-author. - Code Refactor: As a co-author, he participated in the refactoring of the MOE module vllm-project#2150 vllm-project#2706 vllm-project#2867 - Performance Enhancement for RL: Participated as a co-author in the design and development of the solution, contributing to the planning of core capabilities. vllm-project#1547 vllm-project#2120 and so on. So I think he's a great addition to the vLLM Ascend Maintainer team. ---- Chuanyu Qin [@nalinaly](https://github.com/nalinaly) --- The main reason I nominated Qinchuanyu is because he is the initial designer of aclgraph and torch-npu, two key components of vllm-ascend. Considering aclgraph will eventually become the main path for vllm-ascend's graph model, I propose to nominate him. **Sustained and Major Contributions:** In fact, chuanyu actively helped the users/developers of vllm-ascend since Mar 2025 ([vllm-discuss#162](https://discuss.vllm.ai/t/can-ascend-officially-draft-a-documentation-on-the-vllm-ascend-adaptation-for-graph-mode/162/5)), and also helped early users of vllm-ascend understand aclgraph. He provided lots of help in the process of integrating aclgraph with vllm-ascend. **Community Involvement‌:** As speaker, he also presents help users understand aclgraph and torch_npu [《The design philosophy of torch_npu and the high performance principle of aclGraph》](https://github.com/PyTorch-China/pytorch-meetup/blob/main/beijing-2025/%E3%80%905%E3%80%91torch_npu%20%E7%9A%84%E8%AE%BE%E8%AE%A1%E5%93%B2%E5%AD%A6%E4%B8%8E%20aclGraph%20%E9%AB%98%E6%80%A7%E8%83%BD%E5%8E%9F%E7%90%86-%E7%A7%A6%E4%BC%A0%E7%91%9C-0920.pdf) ---- They have activate contribution to vllm-ascend or have rich experience for ascend AI. Welcome! - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: wangxiyuan <[email protected]> Signed-off-by: luolun <[email protected]>

…ject#3001) ### What this PR does / why we need it? Fix issues mentioned in vllm-project#2791 and some minor refactoring. 1. Use Enum instead of string. 2. Avoid setting a new property to forward_context in AscendFusedMoE.forward(). 3. Enabling TokenDispatcherWithMoge. 4. Remove redundant code. ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? Qwen3-30B-A3B/Qwen3-30B-A3B-W8A8/DeepSeek-V3-W4A8-Pruing/deepseek-mtp/pangu-pro-moe-pruing: 1. Enable/Disable EP 2. Aclgraph & eager - vLLM version: v0.10.2 - vLLM main: vllm-project/vllm@9607d5e Signed-off-by: Pr0Wh1teGivee <[email protected]> Co-authored-by: weijinqian0 <[email protected]> Signed-off-by: hwhaokun <[email protected]>

…nalinaly

…nalinaly (vllm-project#3406) I'd like to nominate 4 new maintainers for vllm-ascend: ---- Yizhou Liu [@yiz-liu](https://github.com/yiz-liu) ---- **Review Quality‌**: He has completed [40+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Ayiz-liu) and provided solutions or guides for [10+ issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20commenter%3Ayiz-liu), which includes many quality review like [#issue-3428408401](vllm-project#3002 (comment)), [#discussion_r2224572309](vllm-project#1803 (comment)), [#issuecomment-2982470226](vllm-project#1261 (comment)), [#issuecomment-2903621197](vllm-project#836 (comment)), [#issuecomment-2857678691](vllm-project#778 (comment)). **Sustained and High-Quality Contributions:** He has contributed more than [30+ commits](https://github.com/vllm-project/vllm-ascend/commits?author=yiz-liu) since Mar.2025, especially, aclgraph, DP, and EP related contributions are the main reason why I nominated him. As the owner of aclgraph support, he continuously improves aclgraph stability and performance as well as fixes key bugs. he laid the groundwork for EP-related functionality and delivered multiple foundational improvements **Community involvement:** He has a very good habit of logging issues：vllm-project#1649 and is also very active and involved in [many issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20state%3Aopen%20commenter%3Ayiz-liu%20-author%3Ayiz-liu) to help users resolve issues. ---- Peng Yu [@paulyu12](https://github.com/paulyu12) --- The main reasons for his nomination are his expertise and key contributions to the LORA and sustained and major contributions (initial support/doc/bugfix) around Lora. **Sustained and Major Contributions:** @paulyu12 starts his contribution with [Lora and Mulit-Lora support](vllm-project@697908f) since Apr 2025, he contributed about [10+ commits and bugfixes](vllm-project@697908f) on vllm-ascend. **Review Quality‌ and Community Involvement‌:** He also helped more than 10+ users address [Lora related issues](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Apaulyu12+-author%3Apaulyu12+is%3Aclosed). I believe his addition will further improve vLLM Ascend Lora support. ---- Jinqian Wei [@weijinqian0](https://github.com/weijinqian0) --- The main reasons for his nomination are his key contributions to the RL scene and the high quality of his code reviews. **Review Quality‌:** He has completed [60+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Aweijinqian0+is%3Aopen+-author%3Aweijinqian0) since June. 2025, include [#comment-3284055430](vllm-project#2791 (comment)), [discussion_r2332166704](vllm-project#2817 (comment)), [discussion_r2343289692](vllm-project#2846 (comment)) high quality review. **Sustained and Quality Contributions:** He has Deep understanding of ‌vLLM‌ and ‌vLLM Ascend‌ codebases and solid contributions in RL scene (about [10+ PR merged](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Aweijinqian0+is%3Amerged+) and 10+ PRs merged as co-author. - Code Refactor: As a co-author, he participated in the refactoring of the MOE module vllm-project#2150 vllm-project#2706 vllm-project#2867 - Performance Enhancement for RL: Participated as a co-author in the design and development of the solution, contributing to the planning of core capabilities. vllm-project#1547 vllm-project#2120 and so on. So I think he's a great addition to the vLLM Ascend Maintainer team. ---- Chuanyu Qin [@nalinaly](https://github.com/nalinaly) --- The main reason I nominated Qinchuanyu is because he is the initial designer of aclgraph and torch-npu, two key components of vllm-ascend. Considering aclgraph will eventually become the main path for vllm-ascend's graph model, I propose to nominate him. **Sustained and Major Contributions:** In fact, chuanyu actively helped the users/developers of vllm-ascend since Mar 2025 ([vllm-discuss#162](https://discuss.vllm.ai/t/can-ascend-officially-draft-a-documentation-on-the-vllm-ascend-adaptation-for-graph-mode/162/5)), and also helped early users of vllm-ascend understand aclgraph. He provided lots of help in the process of integrating aclgraph with vllm-ascend. **Community Involvement‌:** As speaker, he also presents help users understand aclgraph and torch_npu [《The design philosophy of torch_npu and the high performance principle of aclGraph》](https://github.com/PyTorch-China/pytorch-meetup/blob/main/beijing-2025/%E3%80%905%E3%80%91torch_npu%20%E7%9A%84%E8%AE%BE%E8%AE%A1%E5%93%B2%E5%AD%A6%E4%B8%8E%20aclGraph%20%E9%AB%98%E6%80%A7%E8%83%BD%E5%8E%9F%E7%90%86-%E7%A7%A6%E4%BC%A0%E7%91%9C-0920.pdf) ---- They have activate contribution to vllm-ascend or have rich experience for ascend AI. Welcome! - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: wangxiyuan <[email protected]> Signed-off-by: hwhaokun <[email protected]>

### What this PR does / why we need it? 1. Replace prepare/finalize operation in fused_moe.py by moe_comm_method.prepare()/finalize() 2. Replace unified_fused_experts by moe_comm_method.fused_experts() in fused_moe.py/w8a8_dynamic.py/w4a8_dynamic.py 3. Add calling _select_moe_comm_method in spec-decode proposers. 4. Currently, w4a8_dynamic does not support gatherep, use all2allv instead. 5. Remove redundant code. ### Does this PR introduce _any_ user-facing change? AllgatherEP switch is disabled in aclgraph/eager mode, just follow the rules in modelrunner_v1._select_moe_comm_method() ### How was this patch tested? e2e & ut - vLLM version: v0.10.2 - vLLM main: vllm-project/vllm@7f6f2c1 Signed-off-by: Pr0Wh1teGivee <[email protected]> Co-authored-by: weijinqian0 <[email protected]> Signed-off-by: nsdie <[email protected]>

…ject#3001) ### What this PR does / why we need it? Fix issues mentioned in vllm-project#2791 and some minor refactoring. 1. Use Enum instead of string. 2. Avoid setting a new property to forward_context in AscendFusedMoE.forward(). 3. Enabling TokenDispatcherWithMoge. 4. Remove redundant code. ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? Qwen3-30B-A3B/Qwen3-30B-A3B-W8A8/DeepSeek-V3-W4A8-Pruing/deepseek-mtp/pangu-pro-moe-pruing: 1. Enable/Disable EP 2. Aclgraph & eager - vLLM version: v0.10.2 - vLLM main: vllm-project/vllm@9607d5e Signed-off-by: Pr0Wh1teGivee <[email protected]> Co-authored-by: weijinqian0 <[email protected]> Signed-off-by: nsdie <[email protected]>

…nalinaly

…nalinaly (vllm-project#3406) I'd like to nominate 4 new maintainers for vllm-ascend: ---- Yizhou Liu [@yiz-liu](https://github.com/yiz-liu) ---- **Review Quality‌**: He has completed [40+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Ayiz-liu) and provided solutions or guides for [10+ issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20commenter%3Ayiz-liu), which includes many quality review like [#issue-3428408401](vllm-project#3002 (comment)), [#discussion_r2224572309](vllm-project#1803 (comment)), [#issuecomment-2982470226](vllm-project#1261 (comment)), [#issuecomment-2903621197](vllm-project#836 (comment)), [#issuecomment-2857678691](vllm-project#778 (comment)). **Sustained and High-Quality Contributions:** He has contributed more than [30+ commits](https://github.com/vllm-project/vllm-ascend/commits?author=yiz-liu) since Mar.2025, especially, aclgraph, DP, and EP related contributions are the main reason why I nominated him. As the owner of aclgraph support, he continuously improves aclgraph stability and performance as well as fixes key bugs. he laid the groundwork for EP-related functionality and delivered multiple foundational improvements **Community involvement:** He has a very good habit of logging issues：vllm-project#1649 and is also very active and involved in [many issues](https://github.com/vllm-project/vllm-ascend/issues?q=is%3Aissue%20state%3Aopen%20commenter%3Ayiz-liu%20-author%3Ayiz-liu) to help users resolve issues. ---- Peng Yu [@paulyu12](https://github.com/paulyu12) --- The main reasons for his nomination are his expertise and key contributions to the LORA and sustained and major contributions (initial support/doc/bugfix) around Lora. **Sustained and Major Contributions:** @paulyu12 starts his contribution with [Lora and Mulit-Lora support](vllm-project@697908f) since Apr 2025, he contributed about [10+ commits and bugfixes](vllm-project@697908f) on vllm-ascend. **Review Quality‌ and Community Involvement‌:** He also helped more than 10+ users address [Lora related issues](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Apaulyu12+-author%3Apaulyu12+is%3Aclosed). I believe his addition will further improve vLLM Ascend Lora support. ---- Jinqian Wei [@weijinqian0](https://github.com/weijinqian0) --- The main reasons for his nomination are his key contributions to the RL scene and the high quality of his code reviews. **Review Quality‌:** He has completed [60+ reviews](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+commenter%3Aweijinqian0+is%3Aopen+-author%3Aweijinqian0) since June. 2025, include [#comment-3284055430](vllm-project#2791 (comment)), [discussion_r2332166704](vllm-project#2817 (comment)), [discussion_r2343289692](vllm-project#2846 (comment)) high quality review. **Sustained and Quality Contributions:** He has Deep understanding of ‌vLLM‌ and ‌vLLM Ascend‌ codebases and solid contributions in RL scene (about [10+ PR merged](https://github.com/vllm-project/vllm-ascend/pulls?q=is%3Apr+author%3Aweijinqian0+is%3Amerged+) and 10+ PRs merged as co-author. - Code Refactor: As a co-author, he participated in the refactoring of the MOE module vllm-project#2150 vllm-project#2706 vllm-project#2867 - Performance Enhancement for RL: Participated as a co-author in the design and development of the solution, contributing to the planning of core capabilities. vllm-project#1547 vllm-project#2120 and so on. So I think he's a great addition to the vLLM Ascend Maintainer team. ---- Chuanyu Qin [@nalinaly](https://github.com/nalinaly) --- The main reason I nominated Qinchuanyu is because he is the initial designer of aclgraph and torch-npu, two key components of vllm-ascend. Considering aclgraph will eventually become the main path for vllm-ascend's graph model, I propose to nominate him. **Sustained and Major Contributions:** In fact, chuanyu actively helped the users/developers of vllm-ascend since Mar 2025 ([vllm-discuss#162](https://discuss.vllm.ai/t/can-ascend-officially-draft-a-documentation-on-the-vllm-ascend-adaptation-for-graph-mode/162/5)), and also helped early users of vllm-ascend understand aclgraph. He provided lots of help in the process of integrating aclgraph with vllm-ascend. **Community Involvement‌:** As speaker, he also presents help users understand aclgraph and torch_npu [《The design philosophy of torch_npu and the high performance principle of aclGraph》](https://github.com/PyTorch-China/pytorch-meetup/blob/main/beijing-2025/%E3%80%905%E3%80%91torch_npu%20%E7%9A%84%E8%AE%BE%E8%AE%A1%E5%93%B2%E5%AD%A6%E4%B8%8E%20aclGraph%20%E9%AB%98%E6%80%A7%E8%83%BD%E5%8E%9F%E7%90%86-%E7%A7%A6%E4%BC%A0%E7%91%9C-0920.pdf) ---- They have activate contribution to vllm-ascend or have rich experience for ascend AI. Welcome! - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 Signed-off-by: wangxiyuan <[email protected]> Signed-off-by: nsdie <[email protected]>

github-actions bot added module:tests module:ops module:core module:quantization labels Sep 6, 2025

gemini-code-assist bot reviewed Sep 6, 2025

View reviewed changes

Pr0Wh1teGivee changed the title ~~refactor common_fused_moe.py~~ refactor fused_moe.py Sep 6, 2025

github-actions bot added the merge-conflicts label Sep 8, 2025

Pr0Wh1teGivee force-pushed the refactor_common_fused_moe branch from a6dc375 to 9ff7384 Compare September 8, 2025 06:56

github-actions bot removed the merge-conflicts label Sep 8, 2025

Pr0Wh1teGivee force-pushed the refactor_common_fused_moe branch from 278a4dc to 4e3a7fe Compare September 8, 2025 12:23

github-actions bot removed module:tests module:core labels Sep 8, 2025

Pr0Wh1teGivee changed the title ~~refactor fused_moe.py~~ [Main] [Refactor] Enable MoECommMethod in Eager Mode Sep 9, 2025

github-actions bot added module:tests module:core merge-conflicts labels Sep 9, 2025

Pr0Wh1teGivee force-pushed the refactor_common_fused_moe branch from 0dd9668 to 495551f Compare September 10, 2025 03:44

github-actions bot removed the merge-conflicts label Sep 10, 2025

Pr0Wh1teGivee force-pushed the refactor_common_fused_moe branch from 495551f to 3bf442d Compare September 10, 2025 06:14

github-actions bot added merge-conflicts and removed merge-conflicts labels Sep 11, 2025

Pr0Wh1teGivee force-pushed the refactor_common_fused_moe branch 4 times, most recently from 87afa53 to 461834a Compare September 11, 2025 09:12

MengqingCao added the ready-for-test start test by label for PR label Sep 13, 2025

weijinqian0 mentioned this pull request Sep 14, 2025

Supports weight prefetching for qwen235B non-quantized and w8a8 models. #2550

Closed

weijinqian0 mentioned this pull request Sep 15, 2025

[Feat] support static EPLB for allgather EP #2366

Closed

Adapt fused_moe to moe_comm_method in eager mode

93cd027

Co-Authored-By: weijinqian0 <[email protected]> Signed-off-by: Pr0Wh1teGivee <[email protected]>

Pr0Wh1teGivee force-pushed the refactor_common_fused_moe branch from b31c231 to 93cd027 Compare September 16, 2025 01:28

wangxiyuan reviewed Sep 16, 2025

View reviewed changes

MengqingCao mentioned this pull request Sep 16, 2025

Register AscendUnquantizedFusedMoEMethod instead of overwrite forward oot #2679

Closed

wangxiyuan reviewed Sep 16, 2025

View reviewed changes

wangxiyuan approved these changes Sep 16, 2025

View reviewed changes

wangxiyuan merged commit 18ca786 into vllm-project:main Sep 16, 2025
16 of 17 checks passed

yiz-liu mentioned this pull request Sep 17, 2025

[RFC]: Refactoring MoE Communication for ACL Graph Compatibility and Performance Optimization #2552

Closed

Pr0Wh1teGivee mentioned this pull request Sep 18, 2025

[Refactor] Adjustments to moe_comm_method selection process #3001

Merged

Yikun mentioned this pull request Sep 20, 2025

[Bug]: Remove outofdate commits to improve perf test #3051

Open

wangxiyuan mentioned this pull request Oct 13, 2025

[Community] Nominate new maintainers: @yiz-liu @paulyu12 @weijinqian0 @nalinaly #3406

Merged

[Main] [Refactor] Enable MoECommMethod in Eager Mode #2791

[Main] [Refactor] Enable MoECommMethod in Eager Mode #2791

Uh oh!

Conversation

Pr0Wh1teGivee commented Sep 6, 2025 • edited by github-actions bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What this PR does / why we need it?

Does this PR introduce any user-facing change?

How was this patch tested?

Uh oh!

github-actions bot commented Sep 6, 2025

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist bot Sep 6, 2025

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented Sep 8, 2025

Uh oh!

github-actions bot commented Sep 9, 2025

Uh oh!

github-actions bot commented Sep 11, 2025

Uh oh!

codecov bot commented Sep 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

weijinqian0 commented Sep 15, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

wangxiyuan left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

wangxiyuan commented Sep 16, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Pr0Wh1teGivee commented Sep 6, 2025 •

edited by github-actions bot

Loading

codecov bot commented Sep 12, 2025 •

edited

Loading