SOURCE: SGI

SGI

March 26, 2015 09:00 ET

SGI Delivers Cutting-Edge Supercomputer to the National Institute for Materials Science

New Supercomputer -- SGI's First Petascale System in Japan -- Boasts 20 Times the Computing Power of Previous System

MILPITAS, CA and TOKYO, JAPAN--(Marketwired - Mar 26, 2015) - SGI Japan, Ltd. part of SGI (NASDAQ: SGI), a global leader in high-performance solutions for compute, data analytics and data management, announced that the National Institute for Materials Science (NIMS) chose to deploy the SGI® ICE™ X distributed memory cluster server. The new supercomputer will be deployed in NIMS' new Numerical Material Simulator, a supercomputer system that will serve as the foundation of the Institute's computational materials science research, enabling NIMS to expand the scale of its numerical models and increase the processing volume of its current research.

National Institute for Materials Science is an Independent Administrative Institution and one of the largest scientific research centers in Japan. NIMS aims to research and develop materials that sustain our societies. The Institute is engaged in research based on computational materials science techniques that aim to clarify the sources of the various properties of materials as well as the research and development of theoretical and computational techniques for the design and prediction of material properties. The availability of the previous Numerical Material Simulator -- utilized for the aforementioned research and development -- was close to saturation due to an expansion in the number of users and to diversification and an increase in the scale of computational processing. In addition, improved computational performance is set to become essential due to expansion in the scale of numerical models and an increase in processing volume. Consequently, the system has been upgraded with the aim of realizing a significant improvement in computational power.

The new SGI ICE X system, deployed in December 2014, is being used for the research and development of cutting-edge computational materials science techniques. These include the analysis and prediction of new properties of nanostructured materials utilizing first principles calculations*, and the statistical thermodynamic analysis of the nanostructure and properties of practical materials. The theoretical computing performance of this new system exceeds one petaflop, and is approximately 20 times faster than the previous system. The large-scale system ranks among the top 10 systems supplied by SGI in the world, and is the first SGI system in Japan to achieve petascale performance.

To respond to the increased requirements NIMS faced, the Institute chose to deploy a SGI ICE X scale-out platform with 1,044 nodes equipped with the latest Intel® Xeon® Processor E5-2600 v3 product family and a total of 25,056 cores, a combined memory of 130 Terabytes (TB), and boasting a theoretical performance of approximately one petaflop -- a 1.4x density increase over the previous system. To reduce heat-generation typically associated with increased performance, the SGI ICE X system is deployed with high energy efficiency standard hot-aisle/cold-aisle installation using innovative, next generation chilled water coils within the system rack. This system allows NIMS to save money on energy costs and reduce resources.

The SGI ICE X cluster is connected to 675 TB of shared physical storage via a high-speed InfiniBand® 4x FDR, leveraging a Lustre File System to facilitate the smooth reading and writing of large volumes of data at high speeds, with a stable file system structure. 

"Intel welcomes the increase in the availability of petascale systems to a wider audience, and looks forward to accelerated research and development through the application of this new system," said Kosuke Hirano, Senior Executive Officer, Business Development, Intel K.K. "Intel will continue to provide advanced solutions in the field of high-performance computing, including processors with increased functionality and performance."

Jorge Titinger, president and CEO of SGI, said, "SGI's work with NIMS is a testament to the success we have seen in enabling large research communities to meet their vast research objectives. SGI is pleased to be working with NIMS and looks forward to collaborating further."

*First principles computing: a calculation technique that uses nanoscale materials simulations. Based on the fundamental equations of quantum mechanics, first principle computing postulates that by analyzing the electron density distribution of a material, its structure, electronic state, and properties can be quantitatively predicted.

About SGI
SGI is a global leader in high performance solutions for compute, data analytics and data management that enable customers to accelerate time to discovery, innovation, and profitability. Visit sgi.com (sgi.com/) for more information.

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

© 2015 Silicon Graphics International Corp. All rights reserved. SGI, the SGI logo, SGI ICE are trademarks or registered trademarks of Silicon Graphics International Corp. or its subsidiaries in the United States and/or other countries. All other trademarks are property of their respective holders. Intel and Xeon are trademarks or registered trademarks of Intel Corporation. All other product and service names mentioned are the trademarks of their respective companies.

Contact Information