RETAIL

MOUNTAIN VIEW, Calif. -- The University of Tokyo's Earthquake Research Institute (ERI) has received a 108-processor SGI(R) Altix(TM) system to be used as a shared supercomputing resource for ERI researchers. The recently delivered supercluster, set to begin production runs later this month, marks the first purchase and delivery of an SGI Altix system worldwide and the first full-scale high-end Linux(R) OS-based parallel processing system in Japan. Launched in January, the SGI Altix family of superclusters features the 64-bit Intel(R) Itanium(R) 2 processor and the Linux operating system. ERI is migrating to the Linux environment by upgrading its current SGI(R) systems to SGI(R) Altix(TM) 3000 family products. The institute, founded over 75 years ago, conducts advanced earthquake and volcano studies, including research on ways to minimize earthquake and volcano damage. It serves as Japan's national research facility to provide opportunities for earthquake and volcano researchers to share information within Japan and internationally. In addition to the SGI Altix 3000 supercluster, which comprises three nodes, SGI Japan has delivered storage systems, terminal hardware, output hardware and network equipment, while also providing total system integration services. For the past four years, the institute has utilized SGI(R) Origin(R) 2000 series systems. The institute chose the SGI Altix 3000 system because it valued the advantages of high scalability and shared memory provided by the SGI(R) NUMAflex(TM) architecture of the Altix 3000 family. Also factoring into the decision was the system's total processing performance and easier system management. SGI Japan's support capabilities, including a successful history with systems engineering and operation support services for the existing Origin 2000 system, were also highly valued. The SGI Altix 3000 family combines the Linux operating system, NUMAflex architecture and Intel Itanium 2 processors into a unique, easy-to-manage environment. Altix 3000 system advances include the ability to scale up to 64 processors and 4TB of memory in a single cluster node and to share memory globally across nodes, potentially involving hundreds of processors. The resulting combination delivers breakthrough performance and reduces the time and resources required to run technical applications.