The HPC lab is equipped with a high-density server to cater to both Windows and Linux environments. The lab not only provides computation resource but also offer a hosting facility to our other departments.
The main component of the Linux environment consists of a High-Performance-Computing cluster. High-Performance Computing (also known as supercomputing) is the process of solving large-scale problems that require substantially more computational horsepower, bandwidth or storage than can be obtained on current workstations and desktop systems.
The "Cheetah" cluster management node is built over Huawei RH2288v3 complimented by 20 compute nodes set up using Huawei XH622. Both management and compute nodes have 2 x Intel Xeon E5-2680v3 (12 Cores) processors backed by 128TB of RAM and approximately 1 TB of storage space each (except for RH2288V3 which has 4TB of storage). One node (Compute-0-20) supports NVidia K40m GPU card to support the GPU computing requirements. Other servers to compliment Linux computing is as below
The windows environment consists of 4 x Intel Xeon E7-4830 v3 (12 Cores) processors with 256 GB RAM and 3TB RAID storage. Multi-core computing is further support with the provision of NVIDIA Tesla K40M. The windows server support up to 10 simultaneous users GUI connections (10 CAL) and allow for a wide variety of research software for computation, development and data analysis. Server backup and researcher data storage are provided with the help of OceanStor 5300 V3 unified storage.
The OceanStor 5300 V3 provide the HPC lab with data storage support. It comes with 2*6 Core CPU,2*16G Cache, and a multitier storage facility. It can work as both a NAS and block storage device. Both Cheetah and Sohrab resources are connected to it, to provide storage and backup facility.
The network interconnects on fibre channel connected to 10 Gbps network switch. The Core Switches Huawei S6720-54c-EI for intercommunication 10 G with listed throughput 2.56 Tbps form the interconnection’s backbone.
