Skip to main content

Run GPU Jobs

In order to run your programs on school GPU, you can pass through our job scheduler service (MaaS) via our command line tool.

Install​

pip install forge-cli --index-url https://gitlab.cri.epita.fr/api/v4/projects/16299/packages/pypi/simple

Once you have run this command, you can use the forge program.

Using the forge program​

You can use the forge program the same way you would use docker run command line.

For instance :

forge run with-gpu -i input_dir -o output_dir reg.undercloud.cri.epita.fr/docker/library/busybox:latest -- ls -la /

Where :

  • input_dir is a directory on your computer that you want to be in /input in the container
  • output_dir is the directory where the contents of /output ends after the execution of the pipeline