The Wall Street Journal
A newly built supercomputer in China appears poised to take the world performance lead, another sign of the country's growing technological prowess that is likely to set off alarms about U.S. competitiveness and national security.
The system was designed by China's National University of Defense Technology and is housed at the National Supercomputing Center in the city of Tianjin. It is part of a new breed that exploits graphics chips more commonly used in playing videogames—supplied by Nvidia Corp.—as well as standard microprocessors from Intel Corp.
Supercomputers are massive machines that help tackle the toughest scientific problems, including simulating commercial products like new drugs as well as defense-related applications such as weapons design and breaking codes. The field has long been led by U.S. technology companies and national laboratories, which operate systems that have consistently topped lists of the fastest machines in the world.
But Nvidia says the new system in Tianjin—which is being formally announced Thursday at an event in China—was able to reach 2.5 petaflops. That is a measure of calculating speed ordinarily translated into a thousand trillion operations per second. It is more than 40% higher than the mark set last June by a system called Jaguar at Oak Ridge National Laboratory that previously stood at No. 1 on a twice-yearly ranking of the 500 fastest supercomputers.
"I don't know of another system that is going to be anywhere near the performance and the power of this machine" in China, said Jack Dongarra, a supercomputer expert on the Oak Ridge research staff who is a professor at the University of Tennessee and recently inspected the system in Tianjin last week. "It is quite impressive."
The development was not altogether unexpected. China placed 24 systems in the so-called Top 500 supercomputer ranking last June; a system called Nebulae, for example, took second place that also used chips from Nvidia and Intel.
But Mr. Dongarra and other researchers said the machine should nevertheless serve as a wake-up call that China is threatening to take the lead in scientific computing—akin to a machine from Japan that took the No. 1 position early in the past decade and triggered increased U.S. investment in the field.
"It's definitely a game-changer in the high performance market," said Mark Seager, chief technology officer for computing at Lawrence Livermore National Laboratory. "This is a phase transition, representative of the shift of economic competitiveness from the West to the East."
Nearly all components of the high-profile Japanese system, called the Earth Simulator, were created in Japan. By contrast, most of the Tianjin system relies on chips from Intel and Nvidia, which are both based in Santa Clara, Calif. So U.S. customers could presumably construct a system with similar performance, noted Horst Simon, deputy lab director at Lawrence Berkeley Lab.
But Mr. Dongarra noted that communications chips inside the machine were proprietary and designed in China, and the country is also working on its own microprocessors.
Moreover, while the Japanese system was a single machine, Tianjin is part of a multi-year strategy by China to develop a range of machines to create a dominant position in both military and commercial applications. "In that sense, I would say this is a much more important event than the Earth Simulator," Mr. Simon said.
The new supercomputer will be operated as an "open access" system, available to other countries outside of China to use for large scale scientific computation, said Ujesh Desai, an Nvidia vice president of product marketing.
It reflects a major design shift to use graphics chips to help accelerate the number-crunching functions most often carried out by so-called x86 chips, which evolved from personal computers and have long dominated supercomputing. Advanced Micro Devices, which makes both graphics chips and x86 microprocessors, is another company besides Nvidia that is promoting the technology shift.
The system was designed by China's National University of Defense Technology and is housed at the National Supercomputing Center in the city of Tianjin. It is part of a new breed that exploits graphics chips more commonly used in playing videogames—supplied by Nvidia Corp.—as well as standard microprocessors from Intel Corp.
Supercomputers are massive machines that help tackle the toughest scientific problems, including simulating commercial products like new drugs as well as defense-related applications such as weapons design and breaking codes. The field has long been led by U.S. technology companies and national laboratories, which operate systems that have consistently topped lists of the fastest machines in the world.
But Nvidia says the new system in Tianjin—which is being formally announced Thursday at an event in China—was able to reach 2.5 petaflops. That is a measure of calculating speed ordinarily translated into a thousand trillion operations per second. It is more than 40% higher than the mark set last June by a system called Jaguar at Oak Ridge National Laboratory that previously stood at No. 1 on a twice-yearly ranking of the 500 fastest supercomputers.
"I don't know of another system that is going to be anywhere near the performance and the power of this machine" in China, said Jack Dongarra, a supercomputer expert on the Oak Ridge research staff who is a professor at the University of Tennessee and recently inspected the system in Tianjin last week. "It is quite impressive."
The development was not altogether unexpected. China placed 24 systems in the so-called Top 500 supercomputer ranking last June; a system called Nebulae, for example, took second place that also used chips from Nvidia and Intel.
But Mr. Dongarra and other researchers said the machine should nevertheless serve as a wake-up call that China is threatening to take the lead in scientific computing—akin to a machine from Japan that took the No. 1 position early in the past decade and triggered increased U.S. investment in the field.
"It's definitely a game-changer in the high performance market," said Mark Seager, chief technology officer for computing at Lawrence Livermore National Laboratory. "This is a phase transition, representative of the shift of economic competitiveness from the West to the East."
Nearly all components of the high-profile Japanese system, called the Earth Simulator, were created in Japan. By contrast, most of the Tianjin system relies on chips from Intel and Nvidia, which are both based in Santa Clara, Calif. So U.S. customers could presumably construct a system with similar performance, noted Horst Simon, deputy lab director at Lawrence Berkeley Lab.
But Mr. Dongarra noted that communications chips inside the machine were proprietary and designed in China, and the country is also working on its own microprocessors.
Moreover, while the Japanese system was a single machine, Tianjin is part of a multi-year strategy by China to develop a range of machines to create a dominant position in both military and commercial applications. "In that sense, I would say this is a much more important event than the Earth Simulator," Mr. Simon said.
The new supercomputer will be operated as an "open access" system, available to other countries outside of China to use for large scale scientific computation, said Ujesh Desai, an Nvidia vice president of product marketing.
It reflects a major design shift to use graphics chips to help accelerate the number-crunching functions most often carried out by so-called x86 chips, which evolved from personal computers and have long dominated supercomputing. Advanced Micro Devices, which makes both graphics chips and x86 microprocessors, is another company besides Nvidia that is promoting the technology shift.