SGI

The Trusted Leader in High Performance Computing

 

Press Release

SGI ICE X Selected by CRIEPI for Electric Power Research

670.9-Teraflop Large-Scale Cluster System with Top-Class Scientific Calculation Performance in Japan

Tokyo, Japan — November 11, 2012 — SGI (NASDAQ: SGI), the trusted leader in technical computing, today announced the installation of the latest high-performance computing (HPC) blade server, known as SGI® ICE™ X, for a new large-scale computer system for scientific calculation at the Central Research Institute of Electric Power Industry (CRIEPI; head office: Chiyoda-ku, Tokyo; president: Masahiro Kakumu). The system boasts a theoretical computing performance of 670.9 teraflops, one of the fastest cluster systems in Japan. It is also the largest and most powerful cluster system in the SGI ICE series supplied by SGI in Japan.

CRIEPI is a research institute founded in 1951 as a joint research institute of electric power companies in order to pursue scientific truth and create technologies that can contribute to the electric power industry and society. CRIEPI's overall aim is to boost efficiency throughout the electric power industry. In recent years, reflecting current societal conditions, research has been focused on areas such as energy, security, advanced techniques in energy production, and construction of next-generation power supply and demand platforms. The enhanced large-scale computer system, in operation since July 2012, has already helped to accelerate this important research.

The new SGI ICE X system will enable more efficient research in areas such as electrical facilities and equipment maintenance; climate change and global environmental trend forecasting for the purpose of disaster mitigation; advanced power generation technologies such as more efficient combustion systems; and structural material design and evaluation research contributing to maintenance and advancement of electric power equipment. This is achieved by using the latest Weather Research and Forecasting (WRF) model and the Vienna Ab initio Simulation Package (VASP) molecular dynamics code which require a very high level of actual calculation performance.

The system is a large scale-out cluster system consisting of 4,032 Intel® Xeon® E5-2670 processors (equivalent to 32,256 cores) with a total memory capacity of 67.5 terabytes. A high-speed InfiniBand® 4x FDR provides interconnection to blade servers with 2,016 nodes (at two processors per node). Storage is provided by a massive 2.88 petabytes of total disk capacity. A water cooling system is used to avoid the need for additional air conditioning, reducing overall power consumption.

The file system, a key component of the scientific calculation system, is the open source Lustre File System that is officially supported by SGI. This parallel file system is highly regarded in the HPC field, and it is being used to construct a fast and stable file system capable of reading and writing large volumes of calculation data.

SGI ICE X is the world's fastest supercomputer¹, factory-integrated and tested to provide high quality and performance and is flexible to fit any workload.

About SGI
SGI, the trusted leader in technical computing is focused on helping customers solve their most demanding business and technology challenges. Visit sgi.com for more information.

Connect with SGI on Twitter (@sgi_corp), YouTube (youtube.com/sgicorp), Facebook (facebook.com/sgiglobal) and LinkedIn.

Media Contact:

SGI Japan
Keiichi Yokoyama
yokoyama@sgi.com
+81-3-5488-1811

Ogilvy Public Relations
Meghan Fintland
415-677-2704
SGImedia@ogilvy.com

© 2012 Silicon Graphics International Corp. SGI, ICE and the SGI logo are trademarks or registered trademarks of Silicon Graphics International Corp or its subsidiaries in the United States and/or other countries. Intel and Xeon are registered trademarks of Intel Corporation. All other trademarks are property of their respective holders.



¹ SGI has submitted a world record SPECmpiL_2007 result of 60.5 for Altix ICE 8400 on the SPEC MPI2007 benchmark. More information on this result is found here: http://www.spec.org/mpi2007/results/mpil2007.html