China's Tianhe-2 Still World's Fastest Supercomputer
That should shake up a top 10 list that has been fairly stagnant over the past couple of years. There was no change in this list for the first nine systems; only the 10th supercomputer—a Cray CS Storm system for an undisclosed U.S. government agency—is new. Tianhe-2 has a max performance of 33.86 petaflops per second, while the Cray system—powered by Intel Xeon E5-2660 v2 chips and Nvidia Tesla K40 GPU accelerators—has a max performance of 3.57 petaflops. The use of accelerators in systems continues to grow. Organizations are increasingly using GPU accelerators from Nvidia or Advanced Micro Devices or x86-based Xeon Phi co-processors from Intel to improve the performance of their systems while holding down power consumption and costs. On the latest list, 75 systems either use GPU accelerators or Intel coprocessors, with 50 of those leveraging Nvidia technology. Three use Radeon GPUs from AMD, while 25 systems are using Xeon Phi coprocessors. Four of the top 10 systems use accelerators—two run Xeon Phis while two more use Nvidia GPUs. Overall, Intel processors power 85.8 percent of all the system on the Top500 list. Hewlett-Packard makes 179—or 36 percent—of the top 500 systems, followed by IBM (153, or 30 percent) and Cray (62 systems, or 12.4 percent). In the June list, HP had 182 systems and IBM 176. The United States was the top supercomputing country, with 231 systems on the Top500 list, though that number declined from 265 on the November 2013 list. "The U.S. is nearing its historical low number on the list," the organizers said in a statement.
European systems grew from 116 in June to 130 now, while the number of supercomputers in Asia fell from 132 to 120, and those in China decreased from 76 to 61. The number of systems in Japan increased from 30 to 32.