THE Department of Energy's Stockpile Stewardship Program is designed to maintain high confidence in the safety, reliability, and performance of the U.S. nuclear weapons stockpile into the indefinite future under a Comprehensive Test Ban Treaty. This objective requires a fundamental understanding of nuclear weapons science and technology through an integrated program incorporating archival nuclear test data, high-fidelity nonnuclear experiments, and advanced computer simulations. The Accelerated Strategic Computing Initiative (ASCI) was developed to provide the simulation capabilities required to meet the stockpile assessment and certification requirements in this uncharted environment. A key element of this initiative is an aggressive plan to acquire computing systems that are substantially beyond those that can be expected simply from market forces. It also represents systems that are (1) at least an order of magnitude greater in speed and memory than any currently in use and (2) three orders of magnitude greater in capability than servers used by most industry and commercial firms and some government laboratories. Unprecedented advances will also be necessary in several areas of computer science and technology to integrate the data transmission, networking, storage, and visualization needs of the ASCI simulation codes.|
To meet this tremendous challenge requires much more than just developing a hardware environment. New simulation codes must be developed that incorporate more fundamental science, a more accurate database describing the properties of materials under nuclear weapon conditions, and new algorithms that take advantage of the massively parallel architectures of the ASCI computers. And the ASCI program's extremely short deadline amplifies the challenge-the target for assembling this high-fidelity three-dimensional simulation capability is the year 2004.
The ASCI approach to accelerating the nation's computer simulation capability reflects the fact that expertise in the many disciplines required to do this job does not necessarily reside within any single laboratory. Many of the important contributions to our nation's technological future must be found within the wider circle of scientific expertise represented in a variety of ASCI partnerships-involving the Sandia, Los Alamos, and Lawrence Livermore national laboratories and featuring collaborations with the commercial computer industry and some of the nation's leading universities.
Here at Livermore, participation in the ASCI program extends across two directorates: the Defense and Nuclear Technologies Directorate and the Computation Directorate. This issue's feature article describes ASCI, a critical element of DOE's Stockpile Stewardship Program.