Interview: China leads world in Exascale computing -- expert

Source: Xinhua| 2017-11-20 15:38:52|Editor: Xiang Bo
Video PlayerClose

by Peter Mertz

DENVER, the United States, Nov. 20 (Xinhua) -- China leads the world in Exascale computing, a technology that could revolutionize human experience on Earth, said Les Cottrell, a lead computer engineer at Stanford University's prestigious National Accelerator Laboratory.

In a recent interview with Xinhua, Cottrell said the term Exascale was coined just three years ago, and shortly thereafter came the announcement of two fastest computers from China that stunned the scientific world.

Exascale computing refers to computing systems capable of at least a billion billion calculations per second.

The two top Chinese supercomputers are Tianhe-2 and the Sunway TaihuLight. Tianhe-2, or Milky Way-2, is capable of speeds of up to 33.86 quadrillion calculations per second, while the Sunway TaihuLight is even faster with amped processing speeds up to 93.01 quadrillion calculations per second.

"China is no longer the poor neighbor," Cottrell said. "In fact, they're leading the world in many areas and competing with companies such as Cisco, Intel and IBM, on a one-to-one basis."

Cottrell spoke to Xinhua at the Super Computer Conference 2017 which was held last week in Denver, in the western U.S. state of Colorado. He once led a team of computer scientists that put the Chinese mainland on the World Wide Web for the very first time -- thanks to a 100,000-U.S.-dollar project financed equally by China and America's Department of Energy (DOE).

Cottrell explained Exascale's breathtaking potential to allow scientists a window into humanity never before imagined.

"Scientists will be able to use Exascale technology to simulate the formation of the Universe," said Cottrell.

Cottrell also cited examples of what the results of massive and high-speed data transfer could yield.

These diverse, computer-based applications can include driverless cars, solving pollution, running highly-efficient cities, and a deep understanding of the human brain.

Cottrell's niche is analyzing data from hard "X-ray diffractions generated from Free Electron Lasers," according to him.

The latter application would be able to analyze the formation of crystals in biological molecules, including vitamins, drugs, proteins and nucleic acids such as the DNA.

It could revolutionize medicine and the understanding of the human brain, and help design pharmaceuticals against diseases and could actually prolong life.

All of this will be made possible by supercomputers, specifically the leap to Exascale technology, which will move data from detectors to supercomputers 1,000 times faster than current.

Predictions on when Exascale will be realized have varied.

At a supercomputing conference in 2009, Computerworld projected Exascale implementation by 2018. In 2012, chipmaker Intel promised to deliver Exascale technology by 2018. A 2013 DOE-funded study puts the Exascale online date "in the 2020 time frame."

Timetables on Exascale's online entrance have varied because data transfer speeds have slowed overall, according to Cottrell.

"Computer speeds increased so quickly over the past 30 years that this speed slowdown was a surprise," he said.

"Today's computers are so fast that it's hard to even remember the old days of analog and vacuum-tube computers," he added.