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

[Bugfix] Disable the dispatch_ffn_combine kernel in MTP path (#4751)

Browse Source 
### What this PR does / why we need it?

This PR is to fix a smoking test failure. Adjust mtp_proposer and
model_runner_v1 to route MTP decoding through the non‑fused MoE
implementation while keeping the overall inference flow unchanged.

- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

Signed-off-by: mojave2 <chenchen145@huawei.com>
Co-authored-by: Mengqing Cao <cmq0113@163.com>
This commit is contained in:
Chen Chen
2025-12-09 22:14:05 +08:00
committed by GitHub
parent cd1c69ee0b
commit 848419d1ba
2 changed files with 12 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
vllm_ascend/spec_decode/mtp_proposer.py
View File

@@ -27,7 +27,8 @@ from vllm.v1.spec_decode.metadata import SpecDecodeMetadata
from vllm.v1.utils import CpuGpuBuffer from vllm.v1.utils import CpuGpuBuffer
from vllm.v1.worker.gpu_input_batch import CachedRequestState, InputBatch from vllm.v1.worker.gpu_input_batch import CachedRequestState, InputBatch
from vllm_ascend.ascend_forward_context import set_ascend_forward_context from vllm_ascend.ascend_forward_context import (MoECommType,
set_ascend_forward_context)
from vllm_ascend.attention.attention_v1 import AscendAttentionState from vllm_ascend.attention.attention_v1 import AscendAttentionState
from vllm_ascend.attention.utils import AscendCommonAttentionMetadata from vllm_ascend.attention.utils import AscendCommonAttentionMetadata
from vllm_ascend.compilation.acl_graph import (ACLGraphWrapper, from vllm_ascend.compilation.acl_graph import (ACLGraphWrapper,
@@ -237,6 +238,9 @@ class MtpProposer(Proposer):
) = self.runner._sync_metadata_across_dp(num_tokens, with_prefill) ) = self.runner._sync_metadata_across_dp(num_tokens, with_prefill)
moe_comm_type = self.runner._select_moe_comm_method(num_tokens) moe_comm_type = self.runner._select_moe_comm_method(num_tokens)
# TODO: remove this after moe_comm_type selection logic is finalized
moe_comm_type = (MoECommType.ALLTOALL if moe_comm_type
== MoECommType.FUSED_ALLTOALL else moe_comm_type)
if skip_attn: if skip_attn:
attn_metadata = None attn_metadata = None

13
vllm_ascend/worker/model_runner_v1.py
View File

@@ -52,8 +52,8 @@ from vllm.distributed.kv_transfer import (get_kv_transfer_group,
has_kv_transfer_group) has_kv_transfer_group)
from vllm.distributed.kv_transfer.kv_connector.v1 import KVConnectorBase_V1 from vllm.distributed.kv_transfer.kv_connector.v1 import KVConnectorBase_V1
from vllm.distributed.parallel_state import (get_dcp_group, get_dp_group, from vllm.distributed.parallel_state import (get_dcp_group, get_dp_group,
get_pcp_group, get_pp_group, get_ep_group, get_pcp_group,
get_tp_group, get_pp_group, get_tp_group,
is_global_first_rank) is_global_first_rank)
from vllm.forward_context import get_forward_context from vllm.forward_context import get_forward_context
from vllm.logger import logger from vllm.logger import logger
@@ -2332,10 +2332,11 @@ class NPUModelRunner(LoRAModelRunnerMixin, ECConnectorModelRunnerMixin):
moe_comm_type = MoECommType.ALLGATHER moe_comm_type = MoECommType.ALLGATHER
elif soc_version in {AscendDeviceType._910_93}: elif soc_version in {AscendDeviceType._910_93}:
moe_comm_type = (MoECommType.MC2 # TODO: drop the EP-size guard when dispatch_ffn_combine supports larger EP sizes
if num_tokens <= self.mc2_tokens_capacity else moe_comm_type = (
MoECommType.FUSED_ALLTOALL if quant_type MoECommType.MC2 if num_tokens <= self.mc2_tokens_capacity else
== "w8a8_dynamic" else MoECommType.ALLTOALL) MoECommType.FUSED_ALLTOALL if quant_type == "w8a8_dynamic"
and get_ep_group().world_size <= 16 else MoECommType.ALLTOALL)
else: else:
raise ValueError(f"Unsupported soc_version: {soc_version}") raise ValueError(f"Unsupported soc_version: {soc_version}")