The Oak Ridge National Laboratory had the fastest supercomputer with a system called Titan, which consisted of Cray and Nvidia technology, according to the twice a year Top500 list.
Specifically, Titan is a Cray XK7 system that hit 17.59 Petaflop/s (quadrillions
of calculations per second) on the Linpack benchmark. Titan features 560,640 processors, including 261,632 Nvidia K20x accelerator cores.
The results will be announced at the SC12 supercomputing conference Monday.
Titan bumped Sequoia, an IBM BlueGene/Q system, to No. 2. Sequoia is installed at the Lawrence Livermore National Laboratory and was No. 1 in June. Sequoia---16.32
Petaflop/s---has 1,572,864 cores. The rest of the top 5 included Fujitsu's K computer in Kobe, Japan, a BlueGene/Q system dubbed Mira at the Argonne National Laboratory and another BlueGene/Q system named JUQUEEN at the Forschungszentrum Juelich in Germany.
Dell also cracked the top 10 list with Stampede, a system based on PowerEdge servers and Intel's Xeon Phi processors. Stampede is installed at the University of Texas in Austin.
By geography, the U.S. had 251 of the top 500 systems with Asia having 123 systems. Europe has 105 systems. There are 72 high performance computing systems installed in China.
Among the notable details:
- Overall, 23 systems have Petaflop/s performance.
- 62 systems are using Nvidia GPUs to accelerate computation.
- 7 are accelerated by Intel's Xeon Phi processors.
- 84.6 percent of systems use processors with six or more cores.
- Intel processors are used in 76 percent of the Top500 systems. AMD is in 12 percent of systems and IBM's Power processors have 10.6 percent of the pie.
- IBM has six of the top 10 systems and 32 entries in the top 100.
- IBM has 192 entries in the Top500 list followed by HP at 149 and Cray at 30.