Remember the great supercomputer wars of the 1980s? Japan was going to beat out the U.S. and determine America's competitive future with the biggest and fastest machines in the world. U.S. Defense Dept. planners saw the big iron as a weapon. Through their vast capacity to do simulations at warp speed, they argued, supercomputers could help Russia and China build better bombs, planes, and precisely targeted missiles. There was a tight ban on exports of the machines to many countries.
Well, guess what. The supercomputer contest is back. But this time it's a race, not a war. The U.S. dropped the lead in the early '90s by decreeing that progress was to come from the commercial sector. In contrast, Japan continued to fund research. Today its Institute of Physical & Chemical Research has the fastest machine in the world -- which is, in part, why Washington reversed its stance. It now supports research into machines like Japan's Earth Simulator and is establishing a supercomputing center at Oak Ridge National Laboratory. The Pentagon's Defense Advance Research Projects Agency sponsors a contest to develop supers for the year 2010.
Washington understands that supers are vital to national competitiveness, as well as security. They are already used for everything from auto design to medical diagnosis to drug discovery. But the grandest use for these systems is discovering new science and technology through simulations. And those discoveries will deliver social and industrial gains in the century ahead.
There's good news here. Speeds have soared over the years, but for large-scale simulations, the need for more speed is infinite. Besides computational brawn, there will be important software breakthroughs ahead, vital in a process where writing the software usually takes much more time than running it. Japan delivered a wake-up call. The U.S. responded. In this race, everyone is likely to benefit over time.