Deep-Learning-Workstation-Setup

Leveraging GPU support with docker compose

Here an example for using docker compose with a pre-built image.

More information on how to access GPUs when using docker compose can be found on the following page Enabling GPU access with Compose.

Running the container

The following command will startup all containers specified in the docker-compose.yml file. Run the command from the same directory where the file is located.

Options for specifying the GPU index or UUID for the docker-compose.yml file. In the docker-compose.yml file a variable is used for specifying the GPU index/UUID.

Running docker compose will per default use the docker-compose.yml file located in the same directory. In the docker-compose.yml file a variable has been defined for assigning the GPU to be used. This is very useful when having a system with multiple GPU that might be shared between multiple users.

1. Specify the GPU index/UUID when when running docker compose up. The command GPU_ID=0 docker compose up will make GPU 0 accessible in the container.

    cd ./examples/nvidia-cuda
    GPU_ID=0 docker compose up
   

2. Specify the GPU in the .env file and this value will be used when running docker compose up.

    # in the './examples/nvidia-cuda/.env' file
    GPU_ID=0
   

   The GPU with index 0 will get passed to the container. If the value of GPU_ID is not set in the .env file, the container will not be able to access the GPU and will run with CPU only.

    docker compose up
   

Removing the container(s)

Run docker compose down for stopping and removing the container.

docker compose down