EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Remove unused row_idx in token_dispatcher (#3442)

Browse Source 
### What this PR does / why we need it?
The `row_idx` parameter is no longer used since
PR[#2689](https://github.com/vllm-project/vllm-ascend/pull/2689), so
remove it across multiple files to remove unnecessary calculations and
parameter passing.

### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
accuracy test passed for Qwen3 235B and DeepSeek V3 671B after this PR.


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

---------

Signed-off-by: CaranLic <740821011@qq.com>
This commit is contained in:
CaranLic
2025-10-15 09:08:31 +08:00
committed by GitHub
parent 3642b64afc
commit 15b2e5c995
11 changed files with 37 additions and 88 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
vllm_ascend/ops/moe/token_dispatcher.py
View File

@@ -61,7 +61,6 @@ class MoETokenDispatcher(ABC):
hidden_states: torch.Tensor,
topk_weights: torch.Tensor,
topk_ids: torch.Tensor,
row_idx: torch.Tensor,
expert_map: Optional[torch.Tensor] = None,
log2phy: Optional[torch.Tensor] = None,
global_redundant_expert_num: int = 0,
@@ -171,7 +170,6 @@ class TokenDispatcherWithMC2(MoETokenDispatcher):
hidden_states: torch.Tensor,
topk_weights: torch.Tensor,
topk_ids: torch.Tensor,
row_idx: torch.Tensor,
expert_map: Optional[torch.Tensor] = None,
log2phy: Optional[torch.Tensor] = None,
global_redundant_expert_num: int = 0,
@@ -330,7 +328,6 @@ class TokenDispatcherWithAllGather(MoETokenDispatcher):
hidden_states: torch.Tensor,
topk_weights: torch.Tensor,
topk_ids: torch.Tensor,
row_idx: torch.Tensor,
expert_map: Optional[torch.Tensor] = None,
log2phy: Optional[torch.Tensor] = None,
global_redundant_expert_num: int = 0,
@@ -422,7 +419,6 @@ class TokenDispatcherWithMoge(MoETokenDispatcher):
hidden_states: torch.Tensor,
topk_weights: torch.Tensor,
topk_ids: torch.Tensor,
row_idx: torch.Tensor,
expert_map: Optional[torch.Tensor] = None,
log2phy: Optional[torch.Tensor] = None,
global_redundant_expert_num: int = 0,
@@ -520,7 +516,6 @@ class TokenDispatcherWithAll2AllV(MoETokenDispatcher):
hidden_states: torch.Tensor,
topk_weights: torch.Tensor,
topk_ids: torch.Tensor,
row_idx: torch.Tensor,
expert_map: Optional[torch.Tensor] = None,
log2phy: Optional[torch.Tensor] = None,
global_redundant_expert_num: int = 0,