Skip to content

[Bugfix] Fix mc2 operator error in aclgraph + ep<16 scenario #2609

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Aug 29, 2025
Merged

[Bugfix] Fix mc2 operator error in aclgraph + ep<16 scenario #2609

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion tests/e2e/multicard/test_qwen3_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,6 +55,7 @@ def test_models_distributed_Qwen3_MOE_TP2_WITH_EP():
tensor_parallel_size=2,
enable_expert_parallel=True,
distributed_executor_backend="mp",
enforce_eager=False,
) as vllm_model:
vllm_model.generate_greedy(example_prompts, max_tokens)

Expand All @@ -71,7 +72,7 @@ def test_models_distributed_Qwen3_MOE_W8A8():
dtype=dtype,
tensor_parallel_size=2,
quantization="ascend",
enforce_eager=False,
enforce_eager=True,
) as vllm_model:
vllm_model.generate_greedy(example_prompts, max_tokens)

Expand Down
3 changes: 2 additions & 1 deletion vllm_ascend/ops/common_fused_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -242,7 +242,8 @@ def forward_impl(self, hidden_states: torch.Tensor,
moe_comm_method_name = forward_context.moe_comm_method_name

# TODO: Can we refactor this logic to model_runner?
if not self.moe_config.use_ep:
# TODO: Adjusted logic to differentiate between A2 and A3, we check ep_size here since mc2 only support ep_size >= 16 on A3 now
if self.moe_config.ep_size < 16:
moe_comm_method_name = "allgathercommimpl"

forward_context.moe_comm_method = getattr(self, moe_comm_method_name)
Expand Down
Loading