China builds world’s fastest supercomputer without U.S. chips
There is no U.S.-made system that comes close to the performance of China's new system, the Sunway TaihuLight. Its theoretical peak performance is 124.5 petaflops, according to the latest biannual release today of the world's Top500 supercomputers. It is the first system to exceed 100 petaflops. A petaflop equals one thousand trillion (one quadrillion) sustained floating-point operations per second.
The most important thing about Sunway TaihuLight may be its microprocessors. In the past, China has relied heavily on U.S. microprocessors in building its supercomputing capacity. The world's next fastest system, China's Tianhe-2, which has a peak performance of 54.9 petaflops, uses Intel Xeon processors.
TaihuLight, which is installed at China's National Supercomputing Center in Wuxi, uses ShenWei CPUs developed by Jiangnan Computing Research Lab in Wuxi. The operating system is a Linux-based Chinese system called Sunway Raise.
The TaihuLight is "very impressive," said Jack Dongarra, a professor of computer science at the University of Tennessee and one of the academic leaders of the Top500 supercomputing list, in a report about the new system.
TaihuLight is running "sizeable applications," which include advanced manufacturing, earth systems modeling, life science and big data applications, said Dongarra. This "shows that the system is capable of running real applications and [is] not just a stunt machine," said Dongarra.
It has been long known that China was developing a 100-plus petaflop system, and it was believed China would turn to U.S. chip technology to reach this performance level. But just over a year ago, in a surprising move, the U.S. banned Intel from supplying Xeon chips to four of China's top supercomputing research centers.
The U.S. initiated this ban because China, it claimed, was using its Tianhe-2 system for nuclear explosive testing activities. The U.S. stopped live nuclear testing in 1992, and now relies on computer simulations. Critics in China suspected the U.S. was acting to slow that nation's supercomputing development efforts.
Four months after the Intel ban, in July 2015, the White House issued an executive order creating a "national strategic computing initiative" with the goal of maintaining an "economic leadership position" in high-performance computing research.
The U.S. order seemed late. China has been steadily building its supercomputing capacity, which included efforts to develop its own microprocessors. It produced a relatively small supercomputer in 2011 that relied on homegrown processors, but its big systems continued to rely on U.S. processors.
There has been nothing secretive about China's intentions. Researchers and analysts have been warning all along that U.S. exascale (an exascale is 1,000 petaflops) development, supercomputing's next big milestone, was lagging.
It's not just China that is racing ahead. Japan and Russia have their own development efforts. Europe is building supercomputers using ARM processors, and, similar to China, wants to decrease its dependency on U.S.-made chips.
China's government last week said it plans to build an exascale system by 2020. The U.S. has targeted 2023.
China now has more supercomputers in the Top500 list than the U.S., said Dongarra. "China has 167 systems on the June 2016 Top500 list compared to 165 systems in the U.S," he said, in an email. Ten years ago, China had 10 systems on the list.
Of all the supercomputers represented on the global list, the sum of the China supercomputers performance (211 petaflops) has exceeded the performance of the supercomputers in the U.S., (173 petaflops) represented on this list. The list doesn't represent the universe of all supercomputers in the U.S. None of the supercomputers used by intelligence agencies, for instance, are represented on this list.
"This is the first time the U.S. has lost the lead," said Dongarra, in the total number of systems on the Top500 list.
China's work is also winning global peer recognition. It's work on TaihuLight has resulted in three submissions selected as finalists for supercomputing's prestigious Gordon Bell Award, named for a pioneer in high-performance computing.
The fastest U.S. supercomputer, number 3 on the Top500 list, is the Titan, a Cray supercomputer at U.S. Dept. of Energy's Oak Ridge National Laboratory with a theoretical peak of about 27 petaflops.
Whether the U.S. chip ban accelerated China's resolve to develop its own microprocessor technology is a question certain to get debate. But what is clear is China's longstanding goal to end reliance on U.S. technology.
"The Chinese were already determined over time to move to an indigenous processor," said Steve Conway, a high performance computing analyst at IDC. "I think the ban accelerates that -- it increases that determination," he said.
HPC has become increasingly important in the economy. Once primarily the domain of big science research, national security and high-end manufacturing such as airplane design, HPC's virtualization and big data analysis capabilities have made it critical in almost every industry. Manufacturers of all sizes, increasingly, are using supercomputers to design products virtually instead of building prototypes. Supercomputer are also used in applications such as fraud detection and big data analysis.
HPC has is now "so strategic that you really don't want to rely on foreign sources for it," said Conway.