Our aim is to get the most out of supercomputing to provide solutions to yours engineering and scientific challenges.
Barcelona, May 21st 2015-- The Donostia International Physics Center Foundation (DIPC) selects the proposal submitted by Omega Peripherals for the supply, installation and set up of its new computer cluster.
The DIPC Foundation, established in 1999, is an institution linked to the University of the Basque Country and one of the international leaders in basic research in the field of material physics. The foundation owns a computer center, the DIPC Computer Center, created to provide its researchers with the computational resources needed to complete their development, testing and research projects. The DIPC Computer Center is equipped with high performance computing systems, scientific support and data storage that allow to perform large simulations and calculations in a fast and efficient way. Currently, the DIPC Computer Center owns and operates eleven clusters. The proposal submitted by Omega Peripherals, a company with a vast experience in the marketing, implementation and maintenance of HPC solutions, storage, and data security, will allow the DIPC to operate a new cluster equipped with Intel Haswell processors that will significantly increase the center’s computing capacity.
Given the need of a new cluster for the DIPC Computer Center, Omega Peripherals presented a solid, reliable, low power consumption solution supplied with the most modern and advanced technologies to provide the greatest performance. Regarding the hardware, the offer is based on ASUS servers supplied by the company Azken Muga that host both the computing nodes and the management node, all of them equipped with 128GB of memory, two 12 core, 2.5GHz processors, and SSD local disks. The scratch parallel filesystem consists of four nodes with BeeGFS. The interconnection of both the compute nodes and the parallel storage is done with an FDR infiniband fabric.
The software installed on the management node is a distribution of the sNow! suite, developed by the HPCNow! team and specifically designed to manage clusters in scientific computing environments. The purpose of sNow! is to provide scientists and engineers with an easy to use and manage system, avoiding the complexities that come with the management of computer clusters.
In order to minimize the impact of calculations in the parallel filesystem the proposal uses BeeGFS on Demand, a function of the BeeGFS software that generates a parallel file system using the local disks of the nodes involved in a computing job.
Physical delivery and installation of the hardware that integrates the solution was jointly conducted by HPCNow! and Azken Muga engineers.
Finally the HPCNow! team was responsible for the set up of the management node, middleware (batch queue manager, math libraries, etc.), and the installation of different versions of the most used applications by the DIPC using EasyBuild for the latter, which greatly facilitates the management of tens or hundreds of applications in a cluster.