China continues to dominate the high end of the Top500 list of the world’s most powerful supercomputers, even as the growth of the computing power on the list seems to be stagnating.
Tianhe-2, run by China’s National University of Defense Technology, clocked 33.86 Pflop/s (quadrillions of calculations per second) for the 43rd edition of the TOP500, released Monday at the International Supercomputing Conference in Leipzig, Germany.
The runner-up in this twice-yearly ranking came in at only half the speed: The U.S. Energy Department’s Titan, a Cray XK7 machine at Oak Ridge National Laboratory, tested out at 17.59 Pflop/s.
Supercomputers perform tasks larger than any single server could handle, such as large-scale calculations for physics research, forecasting the weather, or modeling in high fidelity the design of complex machinery such as airplanes.
This is the third round of the Top500 in which Tianhe-2 has topped the list since its surprise debut a year ago.
The Top 500 list of supercomputers was started in June 1993 to compare the performance of the most powerful computers and show the growth of supercomputing. Participation is voluntary and requires entrants to run the Linpack benchmark, which measures how quickly a system can solve a dense series of linear equations.
Tianhe-2’s continued run on the top of the supercomputing chart is due at least in part to the sluggish growth of the world’s top supercomputers. The machine offered no faster performance than it did a year ago, and nine of the 10 machines on the top of the list also were on previous lists.
Only a new Cray XC30, installed at an undisclosed U.S. government site, made a dent near the top, charting in the 10th position with 3.14 Pflop/s.
Taken together, all 500 systems would produce 274 Pflop/s in supercomputing power, compared with 250 Pflop/s six months ago and 223 Pflop/s in June 2013.
The Top 500 list is compiled by supercomputing experts at the University of Mannheim, Germany; the University of Tennessee, Knoxville; and the Department of Energy’s Lawrence Berkeley National Laboratory.
In addition to ranking supercomputers in terms of Linpack performance, the Top500 also provides insights into the latest ways in which these gargantuan machines are being built.
For example, more supercomputers are using accelerators and co-processors, in addition to the regular processors, to carry out specialized tasks. On this list, 62 systems contained co-processors, up from 53 on the last list in November. Forty-four of these systems use Nvidia co-processors and 17 use Intel Xeon Phi chips.
Despite the victorious Tianhe-2, China still trails behind the U.S. in terms of the number of systems in the Top 500, though it is making gains. China has 76 computers on this list, up from 63 in November. The number of U.S. systems on the current list has shrunk to 233, down from 265 in November.
Intel continues to be the dominant processor provider for these large systems. About 85 percent of the systems on the list use Intel, compared with 8 percent of the systems that use IBM Power processors and 6 percent running AMD Opterons.
Hewlett-Packard continues to be the most widely used supercomputer vendor, having delivered 197 systems on the list, above IBM’s 164 systems and Cray’s 48 systems.