add gatherep select. (#2740)

### What this PR does / why we need it?
add gatherep select.

- vLLM version: v0.10.1.1
- vLLM main:
e599e2c65e

Signed-off-by: wangxiaoxin-sherie <wangxiaoxin7@huawei.com>
Co-authored-by: wangxiaoxin-sherie <wangxiaoxin7@huawei.com>
This commit is contained in:
sherie
2025-09-08 09:15:50 +08:00
committed by GitHub
parent 6666e5265d
commit 2693196ef8
3 changed files with 31 additions and 6 deletions

View File

@@ -45,13 +45,12 @@ def _get_fused_moe_state(ep_size: int, with_prefill: bool,
def get_dispatcher_name(ep_size: int, with_prefill: bool) -> str:
if ep_size == 1:
return "TokenDispatcherWithAllGather"
if ep_size < 16:
elif envs_ascend.VLLM_ENABLE_FUSED_EXPERTS_ALLGATHER_EP and ep_size > 1:
return "TokenDispatcherWithAllGather"
elif ep_size < 16 or with_prefill:
return "TokenDispatcherWithAll2AllV"
if with_prefill:
return "TokenDispatcherWithAll2AllV"
return "TokenDispatcherWithMC2"
else:
return "TokenDispatcherWithMC2"
@contextmanager