The GPU compute cluster offers several hybrid nodes with a balanced CPU, RAM, and GPU ratio. Access may be granted upon request by additional user permissions. The cluster utilizes the workload manager Slurm. You may start your jobs via the login node login-1.gpu.cit-ec.net. For compute tasks it is mandatory to use the Slurm scheduler.
The cluster nodes run on locally installed Ubuntu 22.04. Due to performance issues, the cluster has a separate network file system. Thus the volumes /homes and /vol are not the same folders as on the regular TechFak netboot systems, e. g. compute.techfak.de or the workstations. Therefore on your initial login, you find an empty home folder under /homes/user.
If you have not worked with Slurm, please read our instruction First Steps.
If you have any ideas or you are missing some information on this page, please just create a pull request on the Gitlab repository.
On the Universities Matrix Service join the Channel #citec-gpu:uni-bielefeld.de or send a support request to GPU Support