Theoretically twice as fast, Chinese supercomputer could bump Titan from No. 1 spot

Lights on the Chinese Tianhe-2 supercomputer, which has a theoretical peak that is twice as fast as the Titan supercomputer at Oak Ridge National Laboratory. (Photo courtesy Jack Dongarra)

A new Chinese supercomputer is theoretically twice as fast as the Titan supercomputer at Oak Ridge National Laboratory, and it could bump Titan from the No. 1 spot on the Top 500 list that will come out on June 17, one of the co-authors of the list said Wednesday.

Titan, which reached the No. 1 spot on the semiannual Top 500 list in November, has a theoretical peak of 27 petaflops, or roughly 27,000 trillion calculations per second.

The Chinese supercomputer, Tianhe-2, also known as TH-2 or Milkyway-2, has a theoretical peak of 54.9 petaflops. It also has about twice as much memory as the Titan system, said Jack Dongarra, a Top 500 co-author, University of Tennessee faculty member, and distinguished research staff member in ORNL’s Computer Science and Mathematics Division.

In a Wednesday e-mail, Dongarra said he can’t say who will be No. 1 on the new list, but there is a very strong chance it could be TH-2.

“It’s just a trophy, this No. 1 position,” Dongarra said. “But I think it’s a sign, perhaps a wakeup call for the U.S. Back in 2001, China had no supercomputers—zero. Today, they are No. 2 behind the U.S.”

Computerworld reported that China spent about $290 million on TH-2. Titan is a $100 million machine.

China could stay on top for a few years unless the Japanese, who have been No. 1 in the past, or Europeans catch up first. In a report he wrote after an international forum in China at the end of May, Dongarra said the next large acquisition of a supercomputer for the U.S. Department of Energy will not be until 2015.

Dongarra is a university distinguished professor at the University of Tennessee in the Electrical Engineering and Computer Science Department, and he directs the Innovative Computing Laboratory.

He said the TH-2 is up and running now.

“They are completing their acceptance testing and are beginning to run applications on their system,” Dongarra said. “The machine will be moved from its development site in Changsha at the National University for Defense Technology to its permanent location at the National Supercomputer Center in Guangzhou, China.”

Dongarra said the new supercomputer could be moved to Guangzhou, which is in southwest China, by the end of the year.

He said TH-2 and Titan both use hybrid technology including multicore processors—Intel Ivy Bridge processors for TH-2 and AMD Interlagos for Titan—and accelerators to enhance performance: Intel Xeon Phi co-processors for TH-2 and Nvidia Keplers for Titan.

“Both systems require special attention to programming to take advantage of the accelerators,” Dongarra said. “The Titan has about half a million processors, and the TH-2 has over three million processors. Both machines require special attention when programming them at scale.”

He said TH-2 has a number of features that originated with the Chinese and are unique and interesting, including a certain type of interconnection network, 16-core processor, and the apparent reliability and scalability of the system.

It’s not the first time a system was built with Chinese parts, Dongarra said.

“There was the TH-1a, which used similar Chinese parts created by the same people that made the TH-2, the National University for Defense Technology,” he said. The Tianhe-1A supercomputer won the global title as the world’s fastest in November 2010, Computerworld reported.

Dongarra’s report is available at http://bit.ly/tianhe-2-dongarra-report.

Here is more information he provided about supercomputing in the Wednesday e-mail:

Supercomputing provides capability benefits to a broad range of industries, including energy, pharmaceutical, aircraft, automobile, entertainment, and others. More powerful computing capability allow these diverse industries to more quickly engineer superior new products that could improve a nation’s competitiveness. In addition, there are considerable flow-down benefits that will result from meeting both the hardware and software high-performance computing challenges. These would include enhancements to smaller computer systems and many types of consumer electronics, from smartphones to cameras.

Supercomputers enable simulation—that is, the numerical computations to understand and predict the behavior of scientifically or technologically important systems—and therefore accelerate the pace of innovation. Simulation enables better and more rapid product design. Simulation has already allowed Cummins to build better diesel engines faster and less expensively, Goodyear to design safer tires much more quickly, Boeing to build more fuel-efficient aircraft, and Procter and Gamble to create better materials for home products. Simulation also accelerates the progress of technologies from laboratory to application. Better computers allow better simulations and more confident predictions. The best machines today are 10,000 times faster than those of 15 years ago, and the techniques of simulation for science and national security have been improved.

The use of supercomputers and what’s required to design and produce them is not a U.S. birthright. Without a sustained investment in the U.S. into the technology required to produce these systems, the U.S. could easily lose. Perhaps this is a wakeup call.

Sustaining and more widely exploiting the U.S. competitive advantage in simulation requires concerted efforts toward two distinct goals. First, we must continue to push the limits of hardware and software. Second, to remain competitive globally, U.S. industry must better capture the innovation advantage that simulation offers. But bringing such innovation to large and small firms in diverse industries requires public-private partnerships to access simulation capabilities largely resident in the national laboratories and universities. Of the Top500 supercomputers, half of them are used in industry. Industry gets the importance of simulation and the use of supercomputer to perform those simulations.

More information will be added as it becomes available. This story was last updated at 4:17 p.m.