Successor to world's most powerful supercomputer planned
Connecting state and local government leaders
The director of Japan's Earth Simulator Center is laying plans for an even more powerful successor to the supercomputer under his charge, regarded by many as the world's most powerful computer.
NEWPORT, R.I.'The director of Japan's Earth Simulator Center is laying plans for an even more powerful successor to the supercomputer under his charge, regarded by many as the world's most powerful computer.
Director Tetsuya Sato will be seeking international funding for his new project, he told Government Computer News.
Built by NEC Corp. of Toyko, the 5,100-processor Earth Simulator has demonstrated the ability to execute nearly 36 trillion floating point operations per second (TFLOPS).
When it became operational, the computer inspired worries that the United States is losing dominance in the arena of high- performance computing. Members from 10 agencies, including the Office of Management and Budget, formed the High-End Computing Revitalization Task Force in response (Click for Dec. 15, 2003, GCN story). The House Science Committee held a hearing last summer to evaluate the issue as well.
The most powerful U.S. supercomputer is Los Alamos National Laboratory's ASCI Q, which is less than half as speedy as Japan's, at almost 14 TFLOPS, according to the semi-annual ranking of top computers at Top500.org.
Although Sato would not say how much larger his proposed supercomputer would be, he did say that the problems he is looking to address would require considerably more computational power to solve than the Earth Simulator could provide, perhaps even PFLOP levels of performance. A PFLOP is 1,000 TFLOPS.
Sato this week conducted a presentation of the Earth Simulator's capabilities at the National High Performance Computing and Communications Conference.
Built to address issues raised by the Kyoto Protocol, an international agreement to address global warming, the Earth Simulator models global atmospheric and oceanic conditions as they change over time. Although it can offer a resolution 100 times greater than any other system, some of its results still contain ambiguities, he said.
The new system would merge more types of physical data into the global models, blending microscopic and macroscopic views of different physical influences on the environment. Sato would also want the system to share data with other modeling systems around the world. It would augment, not replace, the Earth Simulator.
Sato said that he felt it would be difficult to acquire the funds needed for such a computer from Japan alone. The Earth Simulator originally cost the country $250 million to build. As a result, Sato would seek international funding for the new project. Also, more countries should be involved in this project because the issues the computer could provide data about are of global concern, Sato said in an interview.
'The problems involve the future of humankind,' he said.
NEXT STORY: DHS analysis chief speaks about plans