Japan Retakes Supercomputing Lead | WPN World People News

After being dethroned in 2004, Japan has retaken the top spot in the most recent release of the Top 500 list of the top-performing supercomputers in the world.

The K Computer at the RIKEN Advanced Institute for Computational Science is the world’s most powerful supercomputer today.

The latest version of the Top 500 list was released yesterday and ranked No. 1 is a system called the K Computer, which is at the RIKEN Advanced Institute for Computational Science in Kobe. The K Computer is capable of performing more than 8 quadrillion calculations per second (petaflop/s).

The Tianhe-1A supercomputer at the National Supercomputing Center in Tianjin, China, is ranked second. This system was ranked No. 1 on the previous Top 500 list, released last November. Also getting bumped down a spot (from No. 2 to No. 3) on the new list is Jaguar, a Cray supercomputer at the U.S. Department of Energy’s (DOE’s) Oak Ridge National Laboratory. These two systems delivered peak measured performances of, respectively, 2.6 petaflop/s and 1.75 petaflop/s.

The list’s rankings are based on a single benchmark, the LINPACK benchmark, which was chosen because it is widely used and performance numbers are available for almost all relevant systems. And while industry experts over the years have noted the shortcomings of the LINPACK benchmark (for that matter, any one benchmark) in measuring a system’s performance, the use of the same benchmark year-to-year makes it easy to make basic comparisons and spot technology trends.

To that end, the Top 500 list is released twice a year (coinciding with annual U.S. and international supercomputing conferences). Rankings are a source of pride for the companies that produce the systems and for the countries using the systems.

For example, with its growing international stature, it is no surprise China increased the number of systems it placed on the list and is now up to 62, making it the No. 2 country. The U.S. is still No. 1, with 256 systems, but that’s down from 274 just six months ago.

Beyond bragging rights, another reason the list generates such high interest is that much of the technology quickly gets incorporated into high-performance computing systems used by businesses. Some of the technology trends in the current list include:

Multi-core processors dominate: Quad-core processors are used in 46.2 percent of the systems, while 42.4 percent of the systems use processors with six or more cores.

Graphics processors as computational accelerators are gaining favor: The two Chinese systems at No. 2 and No. 4 and a Japanese system at No. 5 all use Nvidia GPUs to accelerate computation; a total of 19 systems on the list are using GPU technology in this way.

More cores deliver higher performance: The average concurrency level for systems on this new list is 15,550 cores per system, up from 13,071 six months ago and 10,267 one year ago.

Intel processors dominate: Intel continues to provide the processors for the largest share (77.4 percent) of the Top 500 systems. AMD Opteron family processors are used in 65 systems (13 percent), up from 57. IBM Power processors are used in 45 systems (9 percent), up from 40.

Looking at the entire list, other supercomputing trends emerge. Most notable is the fact that Moore’s Law lives even at this high end of the computing spectrum.

For example, the last system on the new list was listed at position 262 six months ago. That means almost half the systems listed in November got bumped. This turnover rate has steadily increased during the last few lists. And the total combined performance of all 500 systems has grown to 58.88 petaflop/s, compared with 43.7 petaflop/s six months ago and 32.4 petaflop/s one year ago.

Additionally, for the first time, all of the Top 10 systems achieved a performance level in the petaflop/s range. And the entry level to get on the list moved up to the 40.1 teraflop/s mark on the Linpack benchmark, compared with 31.1 teraflop/s six months ago.

Source : Smarter Technology