In what may eventually redefine the term vaporware, scientists studying nuclear weapons have detonated the first "E-bomb." As part of the National Nuclear Security Administration's Stockpile Stewardship Program, which manages the safety, security, and reliability of America's nuclear deterrent, researchers at Los Alamos and Lawrence Livermore national labs detonated two of the largest computer simulations ever.

The simulations ran on the world's fastest supercomputer, ASCI White, and proved it's possible to imitate a nuclear explosion. That gives scientists a way to study the nation's aging nuclear stockpile and build replacement weapon components without violating the nuclear test ban that's been in effect since 1992.

Researchers prepared and revised codes for the Los Alamos, N.M., simulation for about seven months, then needed four months of constant computing by ASCI White to run it. The computation used more than 6.6 million CPU hours, which would take a high-powered home computer more than 750 years to complete. The data consumed was equivalent to 35 times the information available in the Library of Congress.

The Los Alamos simulation was run remotely from Los Alamos on the Livermore White machine, explains Jim Danneskiold, a Los Alamos spokesman. "It worked almost as if [I were] sitting at the local supercomputer," he says.

To gauge the success of the supercomputer modeling, scientists compared the results with the data from a previous underground detonation. The simulations tracked the physics of the actual blast: pressure, temperature, and other data. That requires a massive amount of computational power, Danneskiold says. ASCI White is capable of 12 teraflops, or 1 million computations per second. That's not fast enough for future needs, he says. Los Alamos is installing a 30-teraflop supercomputer that should be running by early next year.

Such computing power isn't limited to weaponry testing. As the technology becomes more widely available, researchers will be able to use supercomputer modeling techniques to speed research in areas ranging from medicine to weather forecasting.