SYSTEMS

Sun Fire x64 Servers Enable Tokyo Tech to Achieve Record Breaking Results in Prestigious Linpack Benchmark -- At Grid Asia 2006, Sun Microsystems today announced that the Tokyo Institute of Technology (Tokyo Tech), one of the world's leading technical institutes, has installed the fastest supercomputer outside the United States. Tokyo Tech's supercomputer, called TSUBAME after the University's symbol bird swallow, is powered by Dual-core AMD Opteron processor-based Sun Fire x64 (x86, 64-bit) servers. This successful installation and benchmark completion marks Sun's largest high performance computing (HPC) win to date, and TSUBAME is expected to be one of the ten largest supercomputers in the world, when the next Top500 list is released in June 2006. The Tokyo Tech TSUBAME is the fastest supercomputer outside of the United States as measured by sustained Linpack performance as of May 2006. TSUBAME is also the largest supercomputer outside the United States and includes more than 21 Terabytes of memory and 1.1 Petabyte of hard disk storage. The system is based on Sun Fire x64 servers with 10,480 AMD Opteron processor cores and Sun storage technologies. Rapid deployment of the system was made possible through NEC's system integration expertise and the use of Sun's N1 management software running on the Solaris 10 Operating System (OS), including N1 System Manager (N1SM) and N1 Grid Engine (N1GE). By using the Sun N1SM and N1GE, the system can support both the Solaris 10 OS and the Linux operating system. Preliminary testing performed at Tokyo Tech clearly demonstrated Sun's ability to deliver the fastest supercomputer outside the United States as measured by sustained performance using the industry standard Linpack Benchmark. The Linpack Benchmark is a highly parallel computing test designed to measure how fast a computer system solves linear equations, a common task in engineering and scientific applications. Tokyo Tech achieved a sustained performance of 38.18 trillion floating point operations per second (TeraFLOPS). Linpack imposes maximum load on a benchmarked machine for up to a day for system of TSUBAME's size, requiring utmost system reliability to properly complete. Similar sized supercomputers often take a year or more to achieve the level of stability required to complete the Linpack test . “This is truly an unprecedented achievement, given that TSUBAME was installed in just three weeks, and it has only been a month since the it went into production,” said Satoshi Matsuoka, Professor in charge of Research Infrastructure at Global Scientific Information and Computing Center, Tokyo Institute of Technology. “And we have only begun to understand the true capability of TSUBAME. We expect further advances in performance with utilization of acceleration, as well as to exploit the other superior aspects of the system, such as very large memory, fat node shared memory architecture with fast interconnect, ultra fast & reliable storage, and most importantly, x86 compatibility with desktop environments.” “Sun is excited to be an integral player in this supercomputing milestone for Tokyo Tech. Our industry-standard 'Galaxy' line of Sun Fire x64 servers, along with the contributions of our partners, have created an HPC solution that surpasses all others outside the U.S. in terms of speed and compute power,” said John Fowler, executive vice president, Network Systems Group at Sun Microsystems. “This triumph is not only significant for Tokyo Tech, but for any company that values swift time-to-market, superior performance and low risk.” Sun partners AMD and NEC play pivotal roles in the development of Asia's largest supercomputer. AMD offers customers the highest performing x64 processor and industry-leading multi-core technology. NEC, the primary systems integrator, is leading the design of the infrastructure as well as the integration of the various applications which will run on the system, based on their extensive experience in building and managing ultra-scale HPC systems. In addition to working with AMD and NEC, Sun’s system also incorporates technology from ClearSpeed Technology Inc., ClusterFS, and Voltaire into the Tokyo Tech system. ClearSpeed Advance Boards installed in the system will be used in the future to accelerate commonly used scientific algorithms; ClusterFS's Lustre parallel file system software allows the servers to communicate with the storage in parallel, speeding access to the ever increasing amounts of scientific data being processed; and Voltaire is supplying high speed multi-protocol Infiniband switches and host card adapters to connect both the servers and storage.