NERSC to Offer 10 Teraflop/s System by Early 2003

BERKELEY, Calif. - The U.S. Department of Energy's National Energy Research Scientific Computing (NERSC) Center has signed a contract with IBM to double the size of NERSC's 3,328-processor RS/6000 SP supercomputer, creating a machine with a peak speed of 10 teraflop/s.

NERSC made the decision to increase the capability of the existing system, rather than purchase an entirely new one, after a thorough review of proposals from a number of leading supercomputer vendors. The new contract, which includes five years of support for the combined system, is valued at about $30 million.

The agreement calls for new equipment to be installed in November with the 6,656-processor IBM system expected to become available to NERSC users by April 2003. This is the largest number of processors ever to be made available on an open production system.

"With the success of DOE's Scientific Discovery through Advanced Computing program, NERSC is seeing an unprecedented level of requests for time on our systems. With this agreement, we have an almost-instant solution to this situation," said Horst Simon, director of the NERSC Center at DOE's Lawrence Berkeley National Laboratory. "During our close working partnership over the past three years, NERSC and IBM have optimized the SP's performance to serve our wide range of users. This new agreement allows us to build on this proven approach and quickly make the expanded system available to the DOE research community. Furthermore the 6,656 processors will give our user community an unprecedented opportunity to explore the scalability of their applications. With 10,000s of processors to be expected on petascale systems by the end of the decade, NERSC scientists will have an early start with this level of parallelism."

NERSC is the flagship supercomputing center for unclassified research sponsored by the Office of Science in DOE. Currently, some 2,100 scientists use NERSC's supercomputers to research problems in combustion, climate modeling, fusion energy, materials science, physics, chemistry and computational biology.

"We had a number of considerations as we looked at our options for increasing NERSC's capabilities, but the bottom line was choosing a system that would have the most cost-effective impact on DOE science," said Bill Kramer, head of NERSC's High-Performance Computing Department and leader of the procurement team. "It was important that we choose a path with the highest likelihood of success and an immediate impact. Not only does this enhancement of our existing IBM SP fulfill those criteria, but it also ensures a very straightforward migration for our large user community."

The new system will include 7.8 terabytes of aggregate memory and a Global Parallel File System with 44 terabytes of storage. The system will be supported by NERSC's High Performance Storage System (HPSS), which provides 2.5 petabytes of archival data storage capacity.

Berkeley Lab is a U.S. Department of Energy national laboratory located in Berkeley, California. It conducts unclassified scientific research and is managed by the University of California. Learn more at http://www.lbl.gov.