It is clear that China and Japan have aggressive road maps to capture leadership of the high-performance computing industry from the United States. The first stop on those road maps occurred in November, when China surged from the pack to claim two of the top three positions on the list of the world’s top 500 supercomputers. This week, Japan surprised the world by blazing into the No. 1 slot with the K Computer — more powerful than the next five systems on the list combined. The Chinese plan to produce their own exascale-class computers — machines with a thousand times the processing power of today’s best supercomputers — before the end of the decade. Japan just signaled its intent to equal or outpace its neighbor. And the two countries hold four of the top five spots on the Top 500 list.
What does leadership in high- performance computing mean? The U.S. Council on Competitiveness has said that “to out-compete is to out-compute.” In its 2008 report “The New Secret Weapon,” the council concluded, “Supercomputing is part of the corporate arsenal to beat rivals by staying one step ahead on the innovation curve. It provides the power to design products and analyze data in ways once unimaginable.”
The potential for the U.S. economy is enormous. Boeing used supercomputers at the Energy Department’s Oak Ridge National Laboratory to accelerate the design of the 787 and 747-8 airliners. In the 1980s, Boeing built and tested 77 wings for the 767. Supercomputer simulations reduced that number to seven for Boeing’s most recent planes. In a 2009 Council on Competitiveness report, Doug Ball, Boeing’s chief engineer, said: “Our work with supercomputers allows us to get a better product out the door faster, which makes us more competitive.”
Navistar Corp. teamed with NASA’s Ames Research Center and the Energy Department’s Lawrence Livermore National Laboratory to use Livermore’s supercomputers to design technologies that improved fuel efficiency in the trucking industry by reducing aerodynamic drag. Simulations demonstrated that simple retrofits could cut annual diesel fuel consumption by 3.4 billion gallons, saving as much as $10 billion a year.
The unique value of exascale computers would also extend to America’s manufacturing sector. Exascale simulation would drive digital design and prototyping of manufactured goods. The time and expense saved would give American industry a critical advantage in an increasingly competitive world market.
China and Japan, of course, are not alone in recognizing the economic value of exascale computing. American computer companies are increasingly marketing their technologies to the growing number of nations, including India, South Korea, Germany, Russia and Australia, that have initiated aggressive, government-funded programs to deliver breakthroughs in supercomputing technology.
These countries realize that exascale-class systems cannot be developed simply by refining existing technologies. Sustained research and development will be needed to overcome a variety of daunting technical challenges.
Among these challenges is a ten-fold reduction in the amount of electricity required to power supercomputers. Without this reduction, future exascale-class systems would use hundreds of megawatts — enough to power a small city — at an annual cost of more than $100 million per system.
An equally important challenge is for supercomputers to be “self-aware,” or instantly compensate for failures. Performing billions of simultaneous tasks will require that exascale machines have a high tolerance for component failure. Integrating these technologies, including low-power components, into consumer products such as cellphones and laptops would be an extraordinary economic opportunity for the nations that develop and bring them to the marketplace.
The Energy Department is addressing these challenges with a national exascale initiative. The endeavor includes an effort to balance development of exascale technologies with efforts to integrate the new technologies into the economy. The initiative builds on partnerships among industry, the Energy Department and its national laboratories, and other agencies such as the National Science Foundation and the Defense Department, whose missions increasingly depend on high- performance computing. The goal of the effort, which will require substantial funding, is to sustain America’s historic leadership in high-performance computing.
As the industrial world emerges from recession, it is imperative that the United States prioritize investments that will sustain our scientific and manufacturing innovation and, through that innovation, our continued leadership in the international economy. The investments in high-performance computing by China, Japan and the rest of the world suggest they have a different future in mind.
Bruce Goodwin is principal associate director of weapons and complex integration at Lawrence Livermore National Laboratory. Thomas Zacharia is deputy laboratory director for science and technology at Oak Ridge National Laboratory.