Supercomputer slowdown: World’s quickest system sees zero new challengers

HPC growth stalls with only 1 new supercomputer cracking top list.

The group that measures the world’s Top 500 most effective supercomputers hasn’t crowned a new champion in more than a year.

Tianhe-2, of China’s National Super Computer Center, overran the top spot in June 2013 with a measured speed of 33.86 petaflop/s, and it held on to #1 in both November 2013 list and the June 2014 list released yesterday.

The follow-up to Tianhe-1A, Tianhe-2 uses Ivy Bridge-based Intel Xeons and Intel Xeon Phi for a total of 3.12 million cores. The computer uses 17,808 kilowatts of ability for 1.9 gigaflop/s per watt and will theoretically hit speeds of up to 54.9 petaflops.

A system maintaining the #1 place three times found in a row isn’t unprecedented: IBM’s US-based Roadrunner, the first petaflop equipment, won three right titles from June 2008 to June 2009. But in the hottest list, the very best nine machines are similar to those from half a year ago. And even underneath of the most notable 500 is seeing significantly less turnover and growth than usual.

“Since its inception in June 1993, the TOP500 list has offered as a consistent measure of the performance growth of supercomputers, since all systems are rated according to performance running the same Linpack benchmark request,” yesterday’s announcement explained. “For the next consecutive list, the entire growth fee of all systems is at a historical low.”

The just new entry in the top 10 “was at number 10-a 3.14 petaflop/s Cray XC30 installed at an undisclosed US federal government site,” the very best 500 job leaders wrote. A petaflop is definitely one quadrillion, or a thousand trillion, calculations per second.

“Things seem to be slowing down”, University of Tennessee professor Jack Dongarra, who created the Linpack benchmarks and helps compile the bi-annual Leading 500 list, told Wired. You may characterize it as maybe a sign that Moore’s Laws is having some concerns.

Some further stats help illustrate the slowdown in growth. This time around, the last system on the Top 500 list once was rated as the 384th quickest system in November 2013. “This represents the cheapest turnover level in the list in 2 decades,” the very best 500 announcement said. Basically, fewer new devices are joining the very best 500: In June 2013, the 500th program had fallen completely from #322.

Between 1994 and 2008, the last system at the top 500 list grew performance by an average of 90 percent every year. Since that time, the #500 program has improved its performance just 55 percent each year.

The combined performance of all 500 systems hit 274 petaflop/s in the most recent list, up from 250 petaflop/s six months ago and 223 petaflop/s twelve months ago. “This increase in installed performance as well exhibits a obvious slowdown in growth compared to the previous long-term trend,” the announcement said.

Supercomputer makers have been boosting speed partly by using co-processors as “accelerators” to handle a few of the work that could otherwise be achieved by CPUs. Sixty-two of the Top 500 systems have co-processors, with 44 of these using Nvidia’s graphics processing units. Nvidia is wishing to make ARM a major area of the supercomputing world by pairing 64-tad ARM server processors using its GPU accelerators.

Vendors could boost supercomputer performance much more by just creating ever bigger systems-but it wouldn’t necessarily get efficient. Companies want to figure out how to effectively create exascale supercomputers, which will be 1,000 times faster than a petaflop per second. Intel today declared brand-new Xeon Phi processors and a far more useful and lower-latency interconnect, calling the brand new architecture “the first practical step towards exascale.”

2 years ago, Intel said 40 to 50 gigaflops of performance per watt is required to hit an exaflop, but that milestone is regarded as at least many years away. The most effective Best 500 supercomputer, which reaches the Tokyo Institute of Technology and uses both Intel CPUs and Nvidia GPUs, can hit 4.5 gigaflops per watt.

Japan plans 130-petaflops China-beating number-crunching supercomputer

Sadly, name of supercomputing monster is the boring “AI Bridging Cloud Infrastructure.”

Japan is reportedly planning to build a 130-petaflops supercomputer costing $173 million (£131 million) that is thanks for completion next season.

Satoshi Sekiguchi, a director-basic at Japan’s ?National Institute of Advanced Industrial Science and Technology, where in fact the computer will be built, told Reuters: “As far as we know, there is little or nothing out there that is as fast.”

Based on the Top 500 blog listing the world’s fastest computers, the existing number-crunching champ is China’s 93-petaflops Sunway TaihuLight, followed by its Tianhe-2, to arrive in 34 petaflops. Japan’s most powerful system right now is normally a 13.5 petaflops model. Overall, Japan gets the fourth-largest quantity of supercomputers in the most notable 500 listing, following the US, China, and Germany.

The UK comes in sixth; the most effective system in the country is normally housed at the Met Workplace, and includes a max performance of 6.8 petaflops.

Like 498 out of the top 500 devices, Japan’s 27 supercomputers in the most notable 500 list all work Linux, and it is highly likely the brand new system will conduct so as well. It is not yet known who’ll construct the machine for the Japanese government-bidding for the task is open up until December 8.

Japan’s new machine will be used in neuro-scientific Artificial Intelligence, which explains its rather boring brand: “AI Bridging Cloud Infrastructure,” or perhaps ABCI. Sekiguchi told Reuters that the machine will also be employed to “tap medical data to develop new products and services and applications.”

Apparently the program is to permit Japan’s corporations to book time in the supercomputer for a fee, so freeing them from the necessity to use US companies like Google and Microsoft.

The investment in the large system is part of a wider proceed to boost Japan’s standing in the wonderful world of technology. Recently, it’s been rather overshadowed by advancements in South Korea and China.

Even though Japan hopes to leap to the most notable of the supercomputer league desk with the brand new ABCI, China is doubtless constructing better machines that may but deprive Japan of this honour.