High Performance Computer (HPC)

The URZ currently operates the following high performance comute (hpc) systems:

sofja - Infiniband cluster with 288 nodes, 32 cores and 256 GB memory per node (since 2021)
t100 - Infiniband cluster with 172 nodes, 16 Haswell cores and 256 GB memory per node (2016 till 2021)
hpc18 - mini HPC with 18 nodes, 2x16-Core AMD Epyc 7351 (Naples) and 256 GB RAM (since 2018)
gpu18 - NVidia DGX-1 with 1 node, 2x 20-Core Intel Xeon E5-2698v4 (Broadwell), 512 GB RAM and 8x NVidia Tesla V100 GPU

Former systems:

Meggie - 8-way QuadOpteron system, SMP with 32 amd64 cores and 256GB Memory (till 2019)
Kautz - SiCortex SC5832, MIPS-CPU cluster with 5832 Cores (till 2018)
Marvel - 32 prozessor shared memory system - HP GS1280 (till 2011)

Furthermore, servers from various institutes are available to a limited extent:

comp2 - 22 nodes Infiniband cluster (till 2021)

Contakt / Account request:

J. Schulenburg (Tel. 58408), E. Lilienblum (Tel. 57379)

Last Modification: 26.01.2022 - Contact Person: IT-Service