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

Delete redundant codes related to communication (#2717)

Browse Source 
### What this PR does / why we need it?
Delete redundant codes related to communication

### Does this PR introduce _any_ user-facing change?
not involve

### How was this patch tested?
not involve

- vLLM version: v0.10.1.1
- vLLM main:
6c7af8110a

---------

Signed-off-by: 刘哲续 <liuzhexu1@huawei.com>
Co-authored-by: 刘哲续 <liuzhexu1@huawei.com>
This commit is contained in:
henryxuxu0716
2025-09-05 09:39:39 +08:00
committed by GitHub
parent 5b3646ab21
commit 51a2aec115
6 changed files with 7 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
tests/ut/ops/test_fused_ops.py
View File

@@ -139,7 +139,6 @@ def mock_dist_env(mocker: MockerFixture):
patch('torch.distributed.all_gather'), \
patch('torch.distributed.all_to_all_single'), \
patch('vllm_ascend.ops.fused_moe.tensor_model_parallel_all_reduce'), \
patch('vllm_ascend.ops.fused_moe.data_parallel_reduce_scatter'), \
patch('vllm.model_executor.layers.fused_moe.config.get_dp_group',
return_value=mock_dp_and_tp_group(mocker)), \
patch('vllm_ascend.ops.fused_moe.get_ascend_config',

2
tests/ut/torchair/ops/test_torchair_fused_moe.py
View File

@@ -66,8 +66,6 @@ def mock_dist_env(mocker: MockerFixture):
patch('torch.distributed.all_to_all_single', return_value=torch.randn(8, 32)), \
patch('vllm_ascend.torchair.ops.torchair_fused_moe.tensor_model_parallel_all_reduce',
return_value=torch.randn(5, 32)), \
patch('vllm_ascend.torchair.ops.torchair_fused_moe.data_parallel_reduce_scatter',
return_value=torch.randn(5, 32)), \
patch('vllm.model_executor.layers.fused_moe.config.get_dp_group',
return_value=mock_dp_and_tp_group(mocker)), \
patch('vllm_ascend.torchair.ops.torchair_fused_moe.get_ascend_config',

25
vllm_ascend/distributed/communication_op.py
View File

@@ -1,25 +0,0 @@
#
# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# This file is a part of the vllm-ascend project.
#
import torch
from vllm.distributed.parallel_state import get_dp_group
def data_parallel_reduce_scatter(input_: torch.Tensor,
dim: int = -1) -> torch.Tensor:
"""Reduce-Scatter the input tensor across data parallel group."""
return get_dp_group().reduce_scatter(input_, dim)

7
vllm_ascend/distributed/moe_comm_method.py
View File

@@ -7,12 +7,11 @@ import torch.nn as nn
import torch_npu
from vllm.distributed import tensor_model_parallel_all_reduce
from vllm.distributed.parallel_state import (
get_tensor_model_parallel_rank, get_tensor_model_parallel_world_size)
get_dp_group, get_tensor_model_parallel_rank,
get_tensor_model_parallel_world_size)
from vllm.forward_context import get_forward_context
from vllm.model_executor.layers.fused_moe import FusedMoEConfig
from vllm_ascend.distributed.communication_op import \
data_parallel_reduce_scatter
from vllm_ascend.distributed.parallel_state import get_mc2_group
from vllm_ascend.utils import AscendSocVersion, get_ascend_soc_version
@@ -147,7 +146,7 @@ class AllGatherCommImpl(MoECommMethod):
When TP size > 1, all-reduce the hidden states to get the final output.
"""
if self.moe_config.dp_size > 1:
hidden_states = data_parallel_reduce_scatter(hidden_states, dim=0)
hidden_states = get_dp_group().reduce_scatter(hidden_states, 0)
hidden_states = hidden_states[:self.num_tokens]
if reduce_results and (self.moe_config.tp_size > 1

6
vllm_ascend/ops/fused_moe.py
View File

@@ -40,8 +40,6 @@ from vllm.model_executor.layers.quantization.base_config import \
from vllm_ascend.ascend_config import get_ascend_config
from vllm_ascend.ascend_forward_context import FusedMoEState
from vllm_ascend.distributed.communication_op import \
data_parallel_reduce_scatter
from vllm_ascend.distributed.parallel_state import get_mc2_group
from vllm_ascend.ops.expert_load_balancer import ExpertLoadBalancer
from vllm_ascend.ops.layers.experts_selector import select_experts
@@ -537,8 +535,8 @@ class AscendFusedMoE(FusedMoE):
final_hidden_states = final_hidden_states[start:end, :]
dispose_tensor(e_hidden_states)
elif fused_moe_state == FusedMoEState.AllGather:
final_hidden_states = data_parallel_reduce_scatter(
e_hidden_states, dim=0)
final_hidden_states = get_dp_group().reduce_scatter(
e_hidden_states, 0)
final_hidden_states = final_hidden_states[:num_tokens]
dispose_tensor(e_hidden_states)
else:

6
vllm_ascend/torchair/ops/torchair_fused_moe.py
View File

@@ -40,8 +40,6 @@ from vllm.model_executor.layers.quantization.base_config import \
from vllm_ascend.ascend_config import get_ascend_config
from vllm_ascend.ascend_forward_context import FusedMoEState
from vllm_ascend.distributed.communication_op import \
data_parallel_reduce_scatter
from vllm_ascend.distributed.parallel_state import get_mc2_group
from vllm_ascend.ops.expert_load_balancer import ExpertLoadBalancer
from vllm_ascend.ops.sequence_parallel import MetadataForPadding
@@ -1269,8 +1267,8 @@ class TorchairAscendFusedMoE(FusedMoE):
final_hidden_states = final_hidden_states[start:end, :]
dispose_tensor(e_hidden_states)
elif fused_moe_state == FusedMoEState.AllGather:
final_hidden_states = data_parallel_reduce_scatter(
e_hidden_states, dim=0)
final_hidden_states = get_dp_group().reduce_scatter(
e_hidden_states, 0)
final_hidden_states = final_hidden_states[:num_tokens]
dispose_tensor(e_hidden_states)
else: