Tier-2 2025 new cluster#

Toward the end of 2025 a new cluster will be added to the KU Leuven/UHasselt’s Tier-2 infrastructure. In line with wICE it will feature thin nodes, large memory nodes, interactive nodes and GPU nodes. The cluster will be operational in Q1 2026.

Note

The new compute nodes will run on Rocky Linux 9. Note that also wICE will migrate to this OS in November. Use the reservation as described to prepare for this.

Hardware details#

  • 40 thin nodes

    • 40 Granite Rapids nodes

      • 2 Intel Xeon 6972P CPUs @2.4 GHz, 96 cores each

      • 768 GiB RAM

      • default memory per core is TBA

      • 960 GB NVMe local disk

      • partitions TBA batch_granite

  • 10 big memory nodes

    • 2 Intel Xeon 6972P CPUs @2.4 GHz, 96 cores each

    • 1536 GiB RAM

    • default memory per core is TBA

    • 3840 GB NVMe local disk

    • partition TBA bigmem

  • 3 GPU nodes

    • 3 nodes with 24 B200 GPUs in total

      • 2 AMD EPYC 9655 CPUs @2.6 GHz (Turin), 96 cores each

      • 1536 GiB RAM

      • default memory per core is TBA

      • 8 NVIDIA B200 SXM6, 192 GiB GDDR, connected with NVLink

      • 960 GB NVMe local disk

      • partition TBA gpu|gpu_b200

  • 2 interactive nodes

    • 2 Intel Xeon 6972P CPUs @2.4 GHz, 96 cores each

    • 768 GiB RAM

    • default memory per core is TBA

    • 2 NVIDIA RTX 5000, 32 GiB GDDR

    • 960 GB SSD local disk

    • partitions TBA interactive

All nodes are interconnected using an Infiniband NDR network. All nodes are connected to the Lustre parallel file system through an Infiniband HDR-100 network.

Similar to the wICE Sapphire Rapids and the H100 nodes, the new cluster is direct liquid cooled.

Storage#

A new parallel filesystem will be installed together with the cluster (IBM Storage Scale, 5.8 PB, connected over NDR). We will assign scratch storage on this new filesystem and also offer project storage (similar to the current staging storage). The compute nodes will also be connected to the existing Lustre filesystem, so all your existing data will be accessible from the new cluster.