docker --runtime=nvidia default로 바꾸기 (GPU 사용 default)

docker container에서 GPU를 사용하기 위해 docker의 runtime을 nvidia로 설정해줘야한다. 하지만 Pycharm과 같은 IDE에서 docker를 사용할 때는 매우 번거롭다. 그러므로 docker 차원에서 runtime을 nvidia로 default 설정해주자.

 

1. nvidia-smi test

docker run --rm ubuntu nvidia-smi

output:

nvidia-smi: command not found

• --runtime=nvidia

docker run --rm --runtime=nvidia ubuntu nvidia-smi

output:

Wed Feb 26 04:19:07 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 430.50       Driver Version: 430.50       CUDA Version: 10.1     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 1070    Off  | 00000000:01:00.0 Off |                  N/A |
|  0%   32C    P8     6W / 166W |     21MiB /  8117MiB |      1%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
+-----------------------------------------------------------------------------+

2. docker daemon.json 파일 수정

sudo vim /etc/docker/daemon.json

before:

{
    "runtimes": {
        "nvidia": {
            "path": "nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}

after:

{
    "default-runtime": "nvidia",
    "runtimes": {
        "nvidia": {
            "path": "nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}

 

 

3. docker 재시작

sudo service docker restart

 

4. --runtime=nvidia 없이 nvidia-smi test

docker run --rm ubuntu nvidia-smi

output:

Wed Feb 26 04:21:52 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 430.50       Driver Version: 430.50       CUDA Version: 10.1     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 1070    Off  | 00000000:01:00.0 Off |                  N/A |
|  0%   32C    P8     6W / 166W |     21MiB /  8117MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
+-----------------------------------------------------------------------------+

 

출처:

http://drmola.com/bbs_free/318738