2
Qwen
提问帖
有回复
K100-AI DCU部署qwen3.8-27B模型报错
2026-08-27 10:23:59
321
### 部署环境 1:宿主机dtk:24.04 2:部署镜像:harbor.sourcefind.cn:5443/dcu/admin/base/custom:vllm0.18.1-ubuntu22.04-dtk26.04-py3.10-20260810-qwen3.8 3: docker启动命令: ```bash docker run -d \ --name vllm-qwen3.8-27b \ --shm-size 128g \ --ipc=host --privileged \ --device /dev/kfd:/dev/kfd \ --device /dev/mkfd:/dev/mkfd \ --device /dev/dri:/dev/dri \ -v /opt/hyhal:/opt/hyhal \ -v /data/LLM_models/models/Qwen--Qwen3.8-27B/snapshots/master:/models/Qwen3.8_27B \ --group-add video \ --cap-add SYS_PTRACE \ --security-opt seccomp:unconfined \ -e HIP_VISIBLE_DEVICES=4,5,6,7 \ -p 16668:8000 \ --restart always \ harbor.sourcefind.cn:5443/dcu/admin/base/custom:vllm0.18.1-ubuntu22.04-dtk26.04-py3.10-20260810-qwen3.8 \ vllm serve /models/Qwen3.8_27B \ --api-key=sk-xxxxxxxxxxxxxxxxxxxxxxxx \ --gpu_memory_utilization=0.88 \ --root-path=/qwen3 \ --served-model-name=qwen3.8_27b \ --max-num-seqs=20 \ --tensor-parallel-size=4 \ --enable-auto-tool-choice \ --reasoning-parser=qwen3 \ --tool-call-parser=qwen3_coder \ --mm-encoder-tp-mode data \ --enable-prefix-caching \ --speculative-config '{"method":"mtp","num_speculative_tokens":3}' ``` ### docker启动报错 ```bash Traceback (most recent call last): File "/usr/local/bin/vllm", line 3, in from vllm.entrypoints.cli.main import main File "/usr/local/lib/python3.10/dist-packages/vllm/__init__.py", line 14, in import vllm.env_override # noqa: F401 File "/usr/local/lib/python3.10/dist-packages/vllm/env_override.py", line 87, in import torch File "/usr/local/lib/python3.10/dist-packages/torch/__init__.py", line 431, in from torch._C import * # noqa: F403 ImportError: librocm_smi64.so.2: cannot open shared object file: No such file or directory Traceback (most recent call last): File "/usr/local/bin/vllm", line 3, in from vllm.entrypoints.cli.main import main File "/usr/local/lib/python3.10/dist-packages/vllm/__init__.py", line 14, in import vllm.env_override # noqa: F401 File "/usr/local/lib/python3.10/dist-packages/vllm/env_override.py", line 87, in import torch File "/usr/local/lib/python3.10/dist-packages/torch/__init__.py", line 431, in from torch._C import * # noqa: F403 ImportError: librocm_smi64.so.2: cannot open shared object file: No such file or directory ```
全部评论(3)
只看楼主
D
DAP后端技术支持
2楼
提供一个线索:宿主机dcu驱动版本要在6.3.3以上
8/31
网
网友56s8vc
回复 @DAP后端技术支持
已解决,需要升级宿主机驱动和dtk到26.04
9/1
J
Jarvis
1楼
可以参考下cookbook上的部署方案:https://github.com/HYGON-AI/inference-cookbook-das/blob/main/docs/model-deployment/vllm/qwen3.8.md
8/28
没有更多内容了
--
楼主
Level.2
0 帖子
楼主热帖
暂无数据