| User Documentation | Dashboard | Detailed Dashboard | Gitea |
|---|
The HPC (High Performance Computing) cluster is a central computation infrastructure at ISD. Using the SLURM cluster engine (same as on the LRZ cluster), you can run jobs, i.e. simple commands, scripts or arrays, on the CPU + GPU cluster. For training of neural networks, one node is equipped with 2 large memory GPUs.
Please consult the user documentation including wiki and example code.
The documentation is found on the Gitea server (login with your FUL account).
Please note: The documentation is work in progress and will be extended over the coming weeks. More detailed SLURM documentations are available elsewhere, e.g. at the Leibniz Rechenzentrum. Most aspects will also apply to our cluster.
Always keep in mind: For a smooth operation of the cluster, it is important that you specify the correct resource requirements of your jobs. If you specify your resources too high, your jobs might block the cluster and unnecessarily delay jobs of other users.
The Linux-based cluster consists of a head node (or "login node", the entry point for users) and compute nodes. The configuration of the nodes is as follows:
| Head Node | Compute Nodes | |
|---|---|---|
| Machines | 1 | 9 |
| CPU(s) | AMD EPYC 16-Core (7302P) | 2x AMD EPYC 24-Core (7352 or 7402) |
| Physical cores | None for compute | 48 per node, 432 in total |
| RAM | 128 GB | 256 GB per node |
| Local scratch | 0 | 1 TB NVMe-SSD |
| Networking | 10 Gbit | 10 Gbit |
| GPUs | - | NVidia Quadro RTX 5000 (16 GB), 5 nodes 2x NVidia Quadro RTX 6000 (24 GB), 1 node# |
The shared cluster storage is located on the "Bigfoot" Storage system.
One node (#) is equipped with two large memory GPUs and can be used for training of neural networks.
The operating system of the cluster is Ubuntu Linux 18.04.4 LTS (HWE). There is only command-line access, no complete graphical user interface.
Software is mostly handled via environment modules or Singularity containers. You can install your own binaries and use your own containers. Please contact the cluster admins to install more complex packages or to deploy software cluster-wide.
To get more info or a cluster user account, please contact ISD IT.