[Update April 30th, 2026]: Please contact us if you want to migrate your IANVS data to the new cluster.

[Update April 28th, 2026]: IANVS will be available until July 2026.

Dear MLU HPC users,

In preparation for the new HPC cluster at the IT-Servicezentrum, IANVS will be decommissioned on May 1st, 2026. Please contact us if you want to migrate your IANVS data to the new cluster. If you do not want to migrate, save your data to an external system. Note that all data on the /home partition is automatically backed up and will be accessible for a year after the decommissioning process (until April 30th, 2027). Please also note that the /scratch partition is not being backed up.

In case you are unable to save your data until May 1st, 2026, please contact us so we can provide you with your data via our backup system. Thank you for understanding.

The new cluster

The new cluster will be set up during May and is planned to go online in June, 2026. The cluster will consist of 16 CPU nodes, 18 GPU32 nodes (each has 4x NVIDIA RTX PRO 6000 with 96 GB VRAM), and 8 GPU64 nodes (each has 4x NVIDIA H200 HVL with 141 GB VRAM). Each node has 2 AMD EPYC Turin 96-Core processors, 1.5 TB RAM, and NVMe SSDs for quick local scratch space. The Lustre filesystem will be used for data persistence.

Software-wise, the cluster will be running Alma Linux 10 with a 6.12 kernel. We will again use SLURM as a scheduler and are planning to offer running jobs in Apptainer/Singularity containers. User software will be installed using Spack.

The cluster is desigend to be modular so that research groups may extend it with their own hardware which will be able to be used by all cluster users after an amount of time. We are currently finalizing the conditions for this process.

Going forward, we will require users to sign an end-user agreement in order to ensure the continued availability and quality of our HPC services. The end-user agreement will be finalized in the coming weeks. There will also be a change in how we handle account and resource allocations:

• Every user will have a small base contingent of resources so they can run basic jobs.
• Larger resource allocations will need to be associated with a project in our resource management system. If you elected to migrate your IANVS data to the new cluster, we will contact you to associate your data with corresponding projects.
• Projects will need to reapply for extensions to their resource allocations every year. In the process of doing so, we will ask project owners (usually research group leaders) to provide us with publications and/or research output.

The project focus is the reason why we are going to start from scratch software-wise. If you need software installed on the new cluster, please use the usual channels detailed in our wiki when the new cluster is online.

Please contact us if you have any questions: via e-mail or phone, (0345) 55-21864.

Best regards,
ITZ HPC Team