Detailed explanation of the process of using GPU in Docker

Take TensorFlow 2.0 as an example

Download tf-gpu

Select the tf version you want to download in docker hub ( note that you should select the one with GPU and py3 )

https://hub.docker.com/r/tensorflow/tensorflow/

like:

docker pull tensorflow/tensorflow:2.0.3-gpu-py3

If the above download times out, you can configure the Tsinghua source.
Or download it using the following command:

docker pull docker.mirrors.ustc.edu.cn/tensorflow/tensorflow:2.0.3-gpu-py3

Build your own image based on the pulled tf-gpu image

The following dockerfile

FROM docker.mirrors.ustc.edu.cn/tensorflow/tensorflow:2.0.3-gpu-py3

RUN ln -sf /usr/share/zoneinfo/Asia/Shanghai /etc/localtime && echo "Asia/Shanghai" > /etc/timezone && \
	pip3 install -i https://pypi.tuna.tsinghua.edu.cn/simple numpy pandas sklearn scipy matplotlib seaborn pyyaml ​​h5py hdfs

RUN pip install deepctr[gpu] -i http://pypi.douban.com/simple --trusted-host pypi.douban.com
RUN pip install keras==2.3.1 -i http://pypi.douban.com/simple --trusted-host pypi.douban.com


# cd /data/wangguisen/ad_ctr
# docker build -t tf-deepctr:1.0 -f ./dk/Dockerfile_base .

Start the image to check whether the GPU is available

docker run --gpus '"device=0"' \
  --rm -it --name ad_ctr \
  -v /data/wangguisen/ad_ctr:/data/ad_ctr \
  tf-deepctr:1.0

Enter nvidia-smi If information about graphics card 0 appears, it means success.

References:

https://blog.csdn.net/weixin_35725559/article/details/112268434

https://zhuanlan.zhihu.com/p/83691871

This is the end of this article about the detailed explanation of how Docker uses GPU. For more information about how Docker uses GPU, please search for previous articles on 123WORDPRESS.COM or continue to browse the following related articles. I hope you will support 123WORDPRESS.COM in the future!