docker用宿主机显卡驱动_Docker中搭建深度学习环境

宿主机环境要求

1.nvidia-driver与cuda

# 查看驱动
nvidia-smi
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 410.78       Driver Version: 410.78       CUDA Version: 10.0     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  Quadro RTX 5000     Off  | 00000000:B3:00.0  On |                  Off |
| 33%   22C    P8     2W / 230W |    414MiB / 16086MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0      2849      G   /usr/lib/xorg/Xorg                            39MiB |
|    0      2961      G   /usr/bin/gnome-shell                          55MiB |
|    0      3813      G   /usr/lib/xorg/Xorg                           170MiB |
|    0      4400      G   /usr/bin/gnome-shell                          65MiB |
|    0     11281      G   ...quest-channel-token=7490845094831916557    82MiB |
+-----------------------------------------------------------------------------+
#查看cuda版本
nvcc -V

上面可以看到显卡型号为 型号Quadro RTX 5000,

2.cudnn

cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2
#define CUDNN_MAJOR 7
#define CUDNN_MINOR 6
#define CUDNN_PATCHLEVEL 4
--
#define CUDNN_VERSION (CUDNN_MAJOR * 1000 + CUDNN_MINOR * 100 + CUDNN_PATCHLEVEL)

#include "driver_types.h"

3.docker >= 19.03

4.nvidia-docker2

5.nvidia-container-runtime (nvidia-docker2默认安装了nvidia-container-runtime)

(base) root@xxribd-Precision-7920-Tower:/home/xxribd/# nvidia-docker version
NVIDIA Docker: 2.2.2
Client: Docker Engine - Community
 Version:           19.03.7
 API version:       1.40
 Go version:        go1.12.17
 Git commit:        7141c199a2
 Built:             Wed Mar  4 01:22:36 2020
 OS/Arch:           linux/amd64
 Experimental:      false

Server: Docker Engine - Community
 Engine:
  Version:          19.03.7
  API version:      1.40 (minimum version 1.12)
  Go version:       go1.12.17
  Git commit:       7141c199a2
  Built:            Wed Mar  4 01:21:08 2020
  OS/Arch:          linux/amd64
  Experimental:     false
 containerd:
  Version:          1.2.13
  GitCommit:        7ad184331fa3e55e52b890ea95e65ba581ae3429
 nvidia:
  Version:          1.0.0-rc10
  GitCommit:        dc9208a3303feef5b3839f4323d9beb36df0a9dd
 docker-init:
  Version:          0.18.0
  GitCommit:        fec3683

注意 cudnn, cuda, nvidia driver的版本一定要对应

配置宿主机需要的环境

安装nvidia-driver

首先查看一下自己的电脑需要怎样的驱动，我们可以先到 htt