Genius hardware#

The KU Leuven / UHasselt Tier-2 cluster Genius has been in production since 2018. Currently only its login nodes and most of its GPU nodes remain in operation.

Hardware details#

  • 16 GPU nodes, 72 GPU devices

    • 14 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|gpu_p100_debug, specific Slurm options apply

    • 2 V100 nodes

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

      • 768 GiB RAM

      • default memory per core is 21000 MiB

      • 8 NVIDIA V100 SXM2@1.5 GHz, 32 GiB GDDR, connected with NVLink

      • 200 GB SSD local disk

      • partition gpu_v100|gpu_v100_long, specific Slurm options apply

The nodes are connected using an Infiniband EDR network (bandwidth 25 Gb/s).