Tianhe-2: China Beats U.S. In Supercomputing

Updated on

The Tianhe-2 or Milky Way-2 is the fastest machine on the earth, which is double in speed compared to that of the US counterparts, which was till now the fastest, reveals a TOP500 list. The Tianhe-2 has been developed by China’s National University of Defense Technology. The fastest computer Milky Way-2 will be fitted at the National Supercomputer Center in Guangzho, towards the end of this year.

Tianhe-2: China Beats U.S. In Supercomputing

The Cray XK7 or Titan, which held the top position previously, was installed at the U.S. Department of Energy’s (DOE) Oak Ridge National Laboratory. The machine has now lost its position to Tianhe-2

Comparison Between Tianhe-2 & Milky Way 2

The Tianhe-2 recorded 33.86 petaflops on the Linpack benchmark, as against Titan’s score of 17.59 petaflops.

The designing of both the machines is very different. Milky Way 2 has 16,000 nodes, each with two Intel Xeon IvyBridge processors and three Xeon Phiprocessors, which make a total of 3,120,000 computing cores.

The architecture of Titan is 261,632 NVIDIA Corporation (NASDAQ:NVDA) Tesla K20x GPU accelerator cores. Titan saves the maximum energy at present of all systems on the list. The machine absorbs a total of 8.21 MW and delivering 2,143 Mflops per Watt.

Other on the list

Sequoia an IBM BlueGene/Q system, which was previously at the number two spot slipped to number three. It is installed in DOE’s Lawrence Livermore National Laboratory. In 2011, first Sequoia was installed and has a speed of 17.17 petaflops on the Linpack benchmark using 1,572,864 cores.

Sequoia is also among the systems which consume least energy. It consumes a total of 7.84 MW and delivering 2,031.6 Mflops per Watt.

The number four spot is bagged by Fujitsu Limited (ADR) (OTCMKTS:FJTSY) (TYO:6702)’s “K computer”, which is installed at the RIKEN Advanced Institute for Computational Science (AICS) in Kobe, Japan. This machine gives the performance of 10.51 petaflops on the Linpack benchmark using 705,024 SPARC64 processing cores.

The number 5 spot is taken by second BlueGene/Q system, Mira, which is fitted at Argonne National Laboratory with performance of 8.59 petaflops on the Linpack benchmark using 786,432 cores.

Other details

Accelerator/Co-processor technology is being used by a total of 54 systems, which is a decline from 62 systems that used this technology in the last list in November 2012. Among these, 39 systems use NVIDIA chips, 3 use ATI Radeon and 11 systems use Intel Xeon Phi.

Collectively the performance of all 500 systems has increased to 223 petaflops, as against 162 petaflops six months ago and 123 petaflops in the previous year.

The results were declared at the International Supercomputing Conference in Leipzig in Germany yesterday.

Leave a Comment