User Tools

Site Tools


chpc:lengau

Lengau Cluster

The main system at the CHPC for high performance computing is a cluster supercomputer called Lengau — cheetah in seTswana. This peta-scale system consists of Dell servers, powered by Intel processors, using FDR InfiniBand by Mellanox and is managed by the Bright Cluster Manager.

The cluster comprises four main parts:

  • 1368 compute nodes
  • 5 large memory nodes
  • 9 GPU nodes
  • 4 PF storage over the Lustre parallel file system

all connected by

  • 56 Gbps FDR Inifniband network interconnect

LENGAU CLUSTER

System Compute Nodes “FAT” Nodes GPU Nodes
CPU Intel Xeon Intel Xeon Intel Xeon
Cores 24 56 36
Speed 2.6 GHz 2.2 GHz 2.2 GHz
RAM 128GiB / 64GiB 1TiB 32GiB
GPU Nvidia V100
GPU RAM 16GiB / 32GiB

Number of Nodes 1368 5 9 Total CPU Cores 32832 280 Total GPU Devices 36

Memory

148.5 TB

5 TB Rpeak 1.307 PFlops Rmax 1.029 PFlops

Interconnect

FDR Infiniband Network

Shared Storage

4 PB Lustre Storage

Launch Date

7 June 2016

7 June 2016

/app/dokuwiki/data/pages/chpc/lengau.txt · Last modified: 2022/06/28 14:24 by wikiadmin