KFUPM’s High Performance Computing (HPC) team worked with a number of international information technology vendors to bring its HPC project to fruition. Microsoft in Saudi Arabia was one of the firms selected to provide consultancy to the project.
“The KFUPM ITC department under the leadership of Dr. Sadiq Sait, formed a full team of highly technical professionals to evaluate different technologies in the market with the goal of creating a state-of-the-art Computational HPC Lab,” said Ahmed Al-Jeshi, Regional Solution Specialist HPC, Microsoft in Saudi Arabia. “The system that was chosen, a dual-boot HPC cluster, shows another leadership milestone for KFUPM in the region.”
KFUPM and Microsoft worked together to identify different user’s needs and demographics for the HPC. It was discovered that there were highly technical users that could use any HPC system but at the same time, there were many users that had a less technical background, who wanted an easy-to-use, reliable and fast HPC system. A decision was reached by KFUPM to install both Microsoft Windows HPC Server 2008 and Linux.
Windows HPC Server 2008 was installed to make use of its innovative interface and features including the out-of-the-box integrated Job Scheduler, MPI Stack, Administration and Deployment tools, and Monitoring and Reporting tools. Also, the partner's ecosystem that Microsoft has built around the hardware vendors as well as the independent software vendors is an added asset.
The main technical details of the KFUPM HPC platform are:
n HPC IBM Cluster 1350 — 12 teraFLOPS, 128 nodes, 1024 Cores, dual boot.
n Master Nodes — 3x Xeon E5405 Quad-Core, 8 GB Ram, 2x 500 GB HD (Mirrored), CDRW/DVD.
n Compute Nodes — 128 nodes (IBM 3550 rack mounted). Each node dual processor, Quad-Core Xeon E5405 (1.6 GHz). 8 GB Ram, 500 GB HD.
n Communication - Interconnect 10GB Ethernet. Uplink 1000-Base-T Gigabit.
n Operating Systems for Compute nodes (Dual Reboot) — Windows HPC Server 2008, Red Hat Linux 5.2.