Sunday, 8 January 2017

Japan Spending $173M On World's Fastest Supercomputer

The international supercomputer race may soon have a new contender: Japan is reportedly eyeing the world's fastest-ever machine.

Reuters this week revealed the country's plans to spend ¥19.5 billion ($173 million) on the construction of a machine capable of making 130 quadrillion calculations per second—or 130 petaflops.
For comparison, the world's current top-ranking supercomputers—Sunway TaihuLight (China), Tianhe-2 (China, pictured), Titan Cray XK7 (US)—boast Linpack marks of 93 petaflops, 34 petaflops, and 24 petaflops, respectively.

In an attempt to win back the industry crown and keep jobs in Japan, the Ministry of Economy, Trade and Industry (METI) will spend billions to construct the most nimble supercomputer in history. The machine's processing power will be available for a fee to organizations focused on driverless cars, robotics, and medical diagnostics.

"As far as we know, there is nothing out there that is as fast," Satoshi Sekiguchi, a director general at Japan's National Institute of Advanced Industrial Science and Technology (AIST), where the supercomputer will be built, told the news site.

Neither METI nor AIST immediately responded to PCMag's requests for comment.
According to the recent 48th TOP500 list, Japan's fastest supercomputers—Fujitsu's Oakforest-PACS and K computer—landed in sixth and seventh; the Fujitsu machines recorded Linpack marks of 13.6 petaflops and 10.5 petaflops, respectively.

China and the US, however, are the countries to beat, claiming 171 systems each in the latest rankings, accounting for two-thirds of the list. Germany comes next with 32 systems, followed by Japan (27), France (20), and the UK (17).

Nvidia, meanwhile, earlier this year unveiled its DGX-1 supercomputer, which CEO Jen-Hsun Huang billed as the world's first deep-learning supercomputer. Using eight Tesla P100 cores, the $129,000 machine will deliver 17 petaflops of peak performance to universities and hospitals engaging in AI research.

No comments:

Post a Comment