A new supercomputer installation in China has rocketed to the top of the twice-annual ranking of the world’s most powerful supercomputers.
In the latest ranking, released Sunday, the Tianjin National Supercomputer Center’s Tianhe-1A system benchmarked a performance of 2.67 petaflops (quadrillion floating-point calculations per second), surpassing the former top achiever, the U.S. Department of Energy (DOE) Oak Ridge Leadership Computing Facility’s Cray XT5 Jaguar system, which clocked in at 1.75 petaflops in this round.
China’s ascendancy to the head of the Top500.org list seems to be sudden: Word of this new system was first reported last month by The New York Times. Its placement ends a six-year run on the top of the list by the U.S., which started when the DOE’s IBM BlueGene/L system stole the top spot from Japan’s Earth Simulator system, built by NEC. The Earth Simulator system itself was the source of much consternation on the part of the U.S. Congress, which set aside money for the DOE to build systems that would re-establish the country’s lead in high-performance computing.
It remains to be seen whether China’s growing prominence on this list will cause a similar reaction. With 42 systems on the list, China has become the second-most-prominent country here, though it still trails by a large margin the U.S., which has 275 systems. Last June, China had only 24 systems on the list.
“Governments around the world are recognizing the deployment of this technology is a prerequisite to sustaining economic competitiveness,” said Dave Turek, vice president of deep computing for IBM. “It lets you do better product designs, basic research, life sciences, fundamental research in materials.”
The DOE does not appear to be standing still. Its Lawrence Livermore National Laboratory plans to unveil the IBM Sequoia system in 2012 that will exceed 20 petaflops. Also that year, the DOE’s Argonne National Lab should have its next-generation IBM Blue Gene supercomputer running, which will offer 10 petaflops.
The TOP500 list is compiled twice a year by researchers at the University of Mannheim, Germany; the DOE’s Lawrence Berkeley National Laboratory; and the University of Tennessee, Knoxville. Participation, which is voluntary, requires that the computer run the Linpack benchmark, a set of routines that solve linear equations.
In this iteration, seven systems achieved a performance of 1 petaflop or better. Three of these systems reside in the U.S., two in China, and one each in France and Japan.
Intel dominates the list in terms of processors. A total of 406 systems are now using Intel chips, though this is down slightly from June, when 408 Intel-based systems were on the list. Intel competitor Advanced Micro Devices has supplied processors for 57 systems. Twenty-eight systems use GPUs (Graphics Processing Units) to help with calculations.
On the system provider side, IBM and Hewlett-Packard are the leaders. IBM has 200 systems on the list and HP has 158 systems.
Joab Jackson covers enterprise software and general technology breaking news for The IDG News Service. Follow Joab on Twitter at @Joab_Jackson. Joab’s e-mail address is Joab_Jackson@idg.com