35cb7b5234
### What this PR does / why we need it? Add dispatch job to leverage jobs to dynamic devices include 2 stage as below: The dispatch job will spend extra about `10s * parallel number + 30s` time to wait other job launch container and release lock. - **Stage 1: Acquire lock** add a dispatch job, this job use lockfile to acquire locks and then get device number dynamically - **Stage 2.1: Launch container with dynamic device** pass the device number via output and start the container job with dynamic device - **Stage 2.2: Release lock** once the job started, release the lock. In the backend, we use multiple path to setup multiple self host runners as load balancer: ``` $ pwd /home/action $ ll | grep actions drwx------ 6 action action 4096 Mar 7 08:55 actions-runner-01 drwx------ 6 action action 4096 Mar 7 08:55 actions-runner-02 drwx------ 6 action action 4096 Mar 7 08:55 actions-runner-03 drwx------ 6 action action 4096 Mar 7 08:56 actions-runner-04 drwx------ 4 action action 4096 Jan 24 22:08 actions-runner-05 drwx------ 4 action action 4096 Jan 24 22:08 actions-runner-06 ``` ``` adduser -G docker action su action pip3 install docker prettytable sudo yum install procmail ``` ### Does this PR introduce _any_ user-facing change? NO ### How was this patch tested? - CI passed - E2E test manully, triggered 3 jobs in parallel: - [1st job](https://github.com/vllm-project/vllm-ascend/actions/runs/13711345757/job/38348309297) dispatch to /dev/davinci2. - [2nd job](https://github.com/vllm-project/vllm-ascend/actions/runs/13711348739/job/38348316250) dispatch to /dev/davinci3 - [3rd job](https://github.com/vllm-project/vllm-ascend/actions/runs/13711351493/job/38348324551) dispatch to /dev/davinci4 Signed-off-by: Yikun Jiang <yikunkero@gmail.com>
63 lines
2.0 KiB
Python
63 lines
2.0 KiB
Python
#
|
|
# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
|
|
# This file is a part of the vllm-ascend project.
|
|
# Adapted from vllm/tests/basic_correctness/test_basic_correctness.py
|
|
# 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.
|
|
#
|
|
"""Compare the short outputs of HF and vLLM when using greedy sampling.
|
|
|
|
Run `pytest tests/test_offline_inference.py`.
|
|
"""
|
|
import os
|
|
|
|
import pytest
|
|
import vllm # noqa: F401
|
|
from conftest import VllmRunner
|
|
|
|
import vllm_ascend # noqa: F401
|
|
|
|
MODELS = [
|
|
"Qwen/Qwen2.5-0.5B-Instruct",
|
|
]
|
|
os.environ["VLLM_USE_MODELSCOPE"] = "True"
|
|
os.environ["PYTORCH_NPU_ALLOC_CONF"] = "max_split_size_mb:256"
|
|
|
|
TARGET_TEST_SUITE = os.environ.get("TARGET_TEST_SUITE", "L4")
|
|
|
|
|
|
@pytest.mark.parametrize("model", MODELS)
|
|
@pytest.mark.parametrize("dtype", ["half", "float16"])
|
|
@pytest.mark.parametrize("max_tokens", [5])
|
|
def test_models(
|
|
model: str,
|
|
dtype: str,
|
|
max_tokens: int,
|
|
) -> None:
|
|
os.environ["VLLM_ATTENTION_BACKEND"] = "ASCEND"
|
|
|
|
# 5042 tokens for gemma2
|
|
# gemma2 has alternating sliding window size of 4096
|
|
# we need a prompt with more than 4096 tokens to test the sliding window
|
|
prompt = "The following numbers of the sequence " + ", ".join(
|
|
str(i) for i in range(1024)) + " are:"
|
|
example_prompts = [prompt]
|
|
|
|
with VllmRunner(model,
|
|
max_model_len=8192,
|
|
dtype=dtype,
|
|
enforce_eager=False,
|
|
gpu_memory_utilization=0.7) as vllm_model:
|
|
vllm_model.generate_greedy(example_prompts, max_tokens)
|