Japan's K computer has set a record for high-performance computing, according to the new Top500 List released Monday by researchers at the University of Tennessee, Lawrence Berkeley National Laboratory and the University of Mannheim in Germany.
Jointly developed by Fujitsu and the RIKEN Advanced Institute for Computational Science in Kobe, Japan, the K computer succeeded in performing 10.5 quadrillion floating-point operations per second (petaflops), which broke the prior record of 8.16 petaflops achieved last June by an earlier configuration of the same machine.
Named after the Japanese word for 10 quadrillion (pronounced "kei"), Japan's supercomputer currently leads the field because of the SPARC64 VIIIfx chips specifically developed by Fujitsu for petaflop-scale computing . Each processor within the K computer's massively interconnected system contains eight cores running at 2 gigahertz, together with 5 MB of shared memory and an embedded memory controller.
The world's most advanced supercomputer, which is four times more powerful than its nearest competitor, is expected to become fully operational in November 2012. In its final configuration, the K computer will integrate 864 equipment racks collectively housing 88,128 interconnected CPUs.
"We truly hope that the K computer's world-class performance will benefit and deliver value to both Japan and the entire world," said Fujitsu President Masami Yamamoto.
Wide Range of Applications
Among other things, scientists already are planning to use Japan's K computer to analyze the behavior of nano-materials. The results of their simulations are expected to aid in the development of next-generation semiconductor building blocks such as nano-wires and carbon nano-tubes featuring lower power consumption requirements.
What's more, the K computer will be able to simulate seismic wave propagation, helping scientists to more accurately predict the extent to which earthquake-impact zones may be subsequently impacted by tsunamis. Furthermore, the petaflop computational power of the new machine will be harnessed to create new architectural designs that make manmade structures more earthquake-resistant.
Other applications include simulations of atmospheric circulation, which may one day enable weather phenomena forecasts that have a significantly higher level of accuracy than do today's hit-and-miss weather reports.
Additionally, pharmaceutical companies will be able to model the requisite molecules for producing more effective drugs -- even as semiconductor companies use the new machine to study the energy conversion efficiencies of next-generation solar cell designs.
"We will continue to work tirelessly to deliver the K computer and its system software by June 2012, and enable it to be used by a large number of researchers throughout the world," Yamamoto said.
Blowing Away the Competition
Initiated in 1993, the Top500 Project ranks the performance of the world's top 500 computing systems in June and November each year, based on the high-Performance LINPACK (HPL) benchmark for distributed-memory computing machines. The project's goal is to accurately track and detect new trends in high-performance computing.
Though still incomplete, Japan's K computer has blown away the competition this year. For example, China's Tianhe-1A system led the pack in 2010 but the machine's performance of 2.57 petaflops per second was only good enough for a second-place ranking in the latest Top500 List.
The top performing supercomputers located in the United States included three Cray machines, SGI's Pleiades supercomputer at NASA's Ames Research Center and IBM's Roadrunner system.
"This is the first time since we began publishing the list back in 1993 that the top 10 systems showed no turnover," said TOP500 Editor Erich Strohmaier.
Posted: 2011-11-14 @ 11:02pm PT