joanrue / nvidia_dockerfiles

Dockerfiles to test different cuda versions

nvidia_dockerfiles

Dockerfiles to test different cuda versions

1 - Create a Dockerfile that starts with the CUDA base image. Check the existing tags here: https://hub.docker.com/r/nvidia/cuda/tags. For example:

nvidia/cuda:11.2.2-base-ubuntu20.04

2 - Build your Docker image by running the following command in the terminal, in the directory where your Dockerfile is located:

sudo docker build -t cuda1122 .

3 - Configure Visual Code Studio for ssh development

• "Open a remote window"
• "Add a new ssh host"
• "ssh [e-mail]"
• click on files logo, open folder
• terminal --> new terminal,
• on the terminal, run a docker container in the background:

sudo docker run -d --name cupy_test_1122 -v /home/ruequera/git/cupy:/root/cupy  --runtime=nvidia  --gpus all cuda1122

• open a bash shell inside a running container:

cupy/
conda init
source .bashrc 
conda create -n cupy python=3.11
conda activate cupy
cd cupy/
pip install -e .
python test_script.py