Genius

Note

The KU Leuven / UHasselt Tier-2 cluster Genius has been in production from 2018 till end of September 2026, at which point the remaining GPU resources were moved to wICE. This page is only for possible future reference.

Hardware details

• thin nodes

  • 86 Skylake nodes

    • 2 Xeon Gold 6140 CPUs@2.3 GHz (Skylake), 18 cores each
      (1 NUMA domain and 1 L3 cache per CPU)

    • 192 GiB RAM (memory bandwidth and latency measurements)

    • 200 GB SSD local disk

    • partition batch|batch_long and interactive

  • 144 Cascadelake nodes

    • 2 Xeon Gold 6240 CPUs@2.6 GHz (Cascadelake), 18 cores each
      (1 NUMA domain and 1 L3 cache per CPU)

    • 192 GiB RAM (memory bandwidth and latency measurements)

    • default memory per core is 5000 MiB

    • 200 GB SSD local disk

    • partition batch|batch_long

  • 4 Naples AMD nodes

    • 2 EPYC 7501 CPUs@2.0 GHz, 32 cores each
      (4 NUMA domains and 8 L3 caches per CPU)

    • 256 GiB RAM

    • 200 GB SSD local disk

    • partition amd

• 10 big memory nodes

  • 2 Xeon Gold 6140 CPUs@2.3 GHz (Skylake), 18 cores each
    (1 NUMA domain and 1 L3 cache per CPU)

  • 768 GiB RAM

  • default memory per core is 21000 MiB

  • 200 GB SSD local disk

  • partition bigmem|bigmem_long

• 1 huge memory node Superdome

  • 8 Xeon Gold 6132 CPUs@2.6 GHz (Skylake), 14 cores each

  • 6 TiB RAM

  • default memory per core is 53500 MiB

  • partition superdome|superdome_long

• 6 GPU nodes, 24 GPU devices

  • 6 P100 nodes

    • 2 Xeon Gold 6140 CPUs@2.3 GHz (Skylake), 18 cores each
      (1 NUMA domain and 1 L3 cache per CPU)

    • 192 GiB RAM

    • default memory per core is 5000 MiB

    • 4 NVIDIA P100 SXM2@1.3 GHz, 16 GiB GDDR, connected with NVLink

    • 200 GB SSD local disk

    • partition gpu_p100|gpu_p100_long

The nodes were connected using an Infiniband EDR network (bandwidth 25 Gb/s). The islands are indicated on the diagram below.