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

Cleanup uesless env (#5270)

Browse Source 
`VLLM_ENABLE_FUSED_EXPERTS_ALLGATHER_EP` is not used anywhere, let's
remove it.
- vLLM version: release/v0.13.0
- vLLM main:
ad32e3e19c

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
wangxiyuan
2025-12-24 22:07:59 +08:00
committed by GitHub
parent 5018f2d8fd
commit fb3d6ca08c
3 changed files with 0 additions and 80 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
.github/workflows/_e2e_test.yaml vendored
View File

@@ -208,7 +208,6 @@ jobs:
pytest -sv --durations=0 tests/e2e/multicard/test_expert_parallel.py pytest -sv --durations=0 tests/e2e/multicard/test_expert_parallel.py
pytest -sv --durations=0 tests/e2e/multicard/test_external_launcher.py pytest -sv --durations=0 tests/e2e/multicard/test_external_launcher.py
pytest -sv --durations=0 tests/e2e/multicard/test_single_request_aclgraph.py pytest -sv --durations=0 tests/e2e/multicard/test_single_request_aclgraph.py
pytest -sv --durations=0 tests/e2e/multicard/test_fused_moe_allgather_ep.py
# torch 2.8 doesn't work with lora, fix me # torch 2.8 doesn't work with lora, fix me
#pytest -sv --durations=0 tests/e2e/multicard/test_ilama_lora_tp2.py #pytest -sv --durations=0 tests/e2e/multicard/test_ilama_lora_tp2.py

74
tests/e2e/multicard/test_fused_moe_allgather_ep.py
View File

@@ -1,74 +0,0 @@
#
# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
# Copyright 2023 The vLLM team.
#
# 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.
#
"""
Execute the inference of fused_moe_allgather_ep and fused_moe_alltoall_ep.
Run 'pytest tests/multicard/test_fused_moe_allgather_ep.py'.
"""
import os
from unittest.mock import patch
import pytest
from modelscope import snapshot_download # type: ignore
from vllm import SamplingParams
from tests.e2e.conftest import VllmRunner
@pytest.mark.skipif(
True,
reason=
"Current disaggregated pd implementation may cause memory pulse, which will cause this test OOM, skip this test until the ringmla is ready "
)
@patch.dict(
os.environ, {
"VLLM_WORKER_MULTIPROC_METHOD": "spawn",
"TASK_QUEUE_ENABLE": "1",
"VLLM_ENABLE_FUSED_EXPERTS_ALLGATHER_EP": "1"
})
def test_deepseek_v3_moe_fused_allgather_ep_tp2():
example_prompts = ["Hello, my name is"]
sampling_params = SamplingParams(max_tokens=100, temperature=0.0)
with VllmRunner(snapshot_download("vllm-ascend/DeepSeek-V3-Pruning"),
tensor_parallel_size=2,
max_model_len=1024,
dtype="auto",
enable_expert_parallel=True) as vllm_model:
vllm_model.generate(example_prompts, sampling_params)
@pytest.mark.skipif(
True,
reason=
"Current disaggregated pd implementation may cause memory pulse, which will cause this test OOM, skip this test until the ringmla is ready "
)
@patch.dict(os.environ, {
"VLLM_WORKER_MULTIPROC_METHOD": "spawn",
"TASK_QUEUE_ENABLE": "1"
})
def test_deepseek_v3_moe_fused_alltoall_ep_tp2():
example_prompts = ["Hello, my name is"]
sampling_params = SamplingParams(max_tokens=100, temperature=0.0)
with VllmRunner(snapshot_download("vllm-ascend/DeepSeek-V3-Pruning"),
tensor_parallel_size=2,
max_model_len=1024,
dtype="auto",
enable_expert_parallel=True) as vllm_model:
vllm_model.generate(example_prompts, sampling_params)

5
vllm_ascend/envs.py
View File

@@ -68,11 +68,6 @@ env_variables: Dict[str, Callable[[], Any]] = {
# that the correct package is installed. # that the correct package is installed.
"VLLM_VERSION": "VLLM_VERSION":
lambda: os.getenv("VLLM_VERSION", None), lambda: os.getenv("VLLM_VERSION", None),
# Whether to enable fused_experts_allgather_ep. MoeInitRoutingV3 and
# GroupedMatmulFinalizeRouting operators are combined to implement EP.
"VLLM_ENABLE_FUSED_EXPERTS_ALLGATHER_EP":
lambda: bool(int(os.getenv("VLLM_ENABLE_FUSED_EXPERTS_ALLGATHER_EP", '0'))
),
# Whether to enable the model execute time observe profile. Disable it when # Whether to enable the model execute time observe profile. Disable it when
# running vllm ascend in production environment. # running vllm ascend in production environment.
"VLLM_ASCEND_MODEL_EXECUTE_TIME_OBSERVE": "VLLM_ASCEND_MODEL_EXECUTE_TIME_OBSERVE":