[docker]运行指定 GPU
Posted adream307
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了[docker]运行指定 GPU相关的知识，希望对你有一定的参考价值。
docker 内运行运行所有的 GPU
docker run --rm --gpus all nvidia/cuda:11.0-base nvidia-smi
输出结果
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 450.57 Driver Version: 450.57 CUDA Version: 11.0 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 GeForce GTX 106... Off | 00000000:01:00.0 On | N/A |
| 23% 48C P0 24W / 130W | 1222MiB / 6075MiB | 1% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
| 1 GeForce GTX 106... Off | 00000000:03:00.0 Off | N/A |
| 48% 33C P8 6W / 130W | 2MiB / 6078MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
+-----------------------------------------------------------------------------+
只运行 1ID 为 1 的 GPU
docker run --rm --gpus "device=1" nvidia/cuda:11.0-base nvidia-smi
输出结果
Mon Sep 14 04:07:29 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 450.57 Driver Version: 450.57 CUDA Version: 11.0 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 GeForce GTX 106... Off | 00000000:03:00.0 Off | N/A |
| 48% 33C P8 6W / 130W | 2MiB / 6078MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
以上是关于[docker]运行指定 GPU的主要内容，如果未能解决你的问题，请参考以下文章
可以在没有 GPU 的情况下运行 nvidia-docker 吗？
sh 在GCP上运行Docker上的tensorflow GPU
docker容器内运行pytorch多gpu报错 RuntimeError: NCCL Error 2: unhandled system error
构建docker环境能够运行自己的GPU服务，能够快速适应不同宿主机的GPU型号操作系统和驱动。linux版本nvidia驱动cudacudnn，docker+conda安装深度学习环境