“In July 2006, these three universities began jointly developing common specifications for each university’s next-generation supercomputer with an eye toward using their supercomputers collaboratively,” said Shunsuke Yoshizawa, director, Regional Marketing, AMD Japan. “The universities aimed to adopt the most advanced and highest performance technology as early as possible, and have engaged in three pillars of openness that make up this new shared specifications experiment: the use of open source hardware architecture, open source system software, and their desire to make the supercomputers widely available amongst the three institutions. The performance, scalability and advance floating point processing capability of the Quad-Core AMD Opteron processor easily matched their requirements.”
The University of Tokyo supercomputer system is comprised of 952 nodes of the Hitachi HA8000-tc/RS425 technical server with four Quad-Core AMD Opteron processors per node. The system as designed achieves a theoretical peak performance of approximately 140 teraflops. A teraflop or “tflop” is the computing power required to process one trillion floating point operations per second. This theoretical peak performance is the fastest in Japan at the time of launch.
The University of Tsukuba’s system was built by Cray Japan Inc. and Sumisho Computer Systems and is based on Appro International’s Xtreme-X Supercomputer, while the Kyoto University’s system was built around the Fujitsu HX600 HPC server. Both supercomputers are equipped with four Quad-Core AMD Opteron processors per node. The University of Tsukuba’s system can achieve a theoretical peak performance of approximately 95 tflops, while the Kyoto University system tops out at a theoretical peak performance of approximately 61 tflops.
The T2K supercomputers will provide these universities and their researchers with the outstanding system performance and advanced floating point processing enabled by Quad-core AMD Opteron processors. The selection of AMD technology for these highly advanced systems can also provide the universities with much-needed energy efficiency in the form of performance-per-watt and innovative processor-level power management features such as AMD CoolCore™ Technology and Independent Dynamic Core Technology.
The synergy of advanced research expertise in computer science developed over many years at these three universities and the computing power of the new AMD processor-based T2K supercomputers is expected to help address increasingly diverse needs in industry and academia including the large-scale scientific calculations for researching subatomic particles and nuclear energy, astronomy, climate modeling and weather forecasting, and genetics and biomedical advancements, among many others. Additional information about T2K can be found at http://www.open-supercomputer.org.