Genius decommissioning

Most remaining parts of Genius will be decommissioned when we start the pilot phases for the new Tier-2 cluster. This should happen sometime between the third week and the fourth week of March.

Note

Starting from the second week of March, many nodes from the batch* and interactive partitions on Genius will already be blocked to allow large-scale benchmarks to run on the new cluster. You will still be able to send jobs to those Genius partitions, but they may spend more time in the queue because of the reduced capacity.

The decommissioning will consist of retiring the following hardware:

• all Cascadelake CPU nodes (batch* partitions)

• all Skylake CPU nodes (interactive and bigmem* partitions)

• all nodes in the dedicated* partitions

• Superdome (superdome* partitions)

If you are still using the Genius CPU partitions, be prepared to switch to the wICE CPU partitions since the final decommissioning of this hardware is drawing near (towards the end of March, as mentioned above).

The P100 and V100 GPU nodes will be kept online for a longer period (gpu_p100* and gpu_v100* partitions). If your compute tasks rely on these older GPUs because they cannot efficiently use more powerful devices, you are welcome to reach out to us at mailto:hpcinfo@kuleuven.be (in case you haven’t been contacted by us already).

Note that the Tier-2 login nodes will remain in service until the end of 2026.