- Linux
- Docker-CE 19.03 or later
- nvidia-docker (Can be neglected for serial or multicore versions)
In the directory where the Dockerfile is located run: docker build -t artss_docker .
It is not necessary to recreate the Docker image each time. It has to be created initially, and when a change is submitted to the Docker file.
Navigate to the ARTSS main directory (where compile.sh is located).
Then run: docker run --gpus all -it --rm -v $(pwd):/host_pwd -w /host_pwd artss_docker
inside the container run: nvidia-smi to make sure you have access to all the GPUs needed
then run compile.sh [OPTIONS] as you normally would. [OPTIONS] could be for example -c 10 -d -g
Navigate to the ARTSS main directory (where compile.sh is located).
Then run: docker run -it --rm -v $(pwd):/host_pwd -w /host_pwd artss_docker
then run compile.sh [OPTIONS]. Don't use any GPU flags, since they will not work here.
After compiling, ARTSS can be used as usual.
exit-> Leave containerdocker container ls-> List all running Docker containersdocker ps -a-> List all Docker containersdocker image ls-> List all Docker imagesdocker container rm [tag]-> Removes Docker container with tag [tag]docker image rm [tag]-> Removes Docker image with tag [tag]
More Docker commands: https://docs.docker.com/engine/reference/commandline/cli/
The Docker image can be built with ./compile.sh --docker-build
To build ARTSS inside the container use --docker-run or --docker-run-cpu. To change the hostname of the Docker add the option --docker-hostname.
The Docker commands can be combined with each other.