SGI Fed Wins Livermore Supercomputing Work

SGI Federal Inc. and Linux NetworX Inc. won a $2 million contract to provide a Linux supercomputing cluster to the National Nuclear Security Administration.

SGI Federal Inc., along with cluster solution provider Linux NetworX Inc., Sandy, Utah, won a $2 million contract to provide a Linux supercomputing cluster to the Department of Energy's National Nuclear Security Administration, the companies announced Aug. 28.

The companies also announced a vendor relationship to pursue additional opportunities.

The clusters will be used for the administration's Accelerated Strategic Computing Initiative, a 10-year program to maintain the safety and reliability of the U.S. nuclear stockpile by simulating the aging and operation of nuclear weapons in three dimensions.

They will be managed by the Lawrence Livermore National Laboratory, a Department of Energy lab operated by the University of California.

"The clustering expertise and management tools delivered by Linux NetworX along with SGI Federal's project management background and support infrastructure created a very powerful and cost-effective solution to meet our requirements," said Mark Seager, an assistant department head for terascale systems at Lawrence Livermore.

An alternative to traditional supercomputers, clustering technology links multiple servers to form a unified system that can match supercomputer performance at a fraction of the cost.

For this project, 472 Pentium 4 processors will be used for three parallel capacity resource clusters with a theoretical peak performance of 857 billion operations per second.

As a result of the win, SGI Federal, a Silver Spring, Md.-based, wholly owned subsidiary of Silicon Graphics Inc., Mountain View, Calif., and Linux NetworX have established a joint vendor relationship to bid on large cluster contracts.

In related news Aug. 15, the national lab fully activated IBM's ASCI White supercomputer.

This $110 million IBM machine, also used for stockpile monitoring, is capable of 12.3 trillion operations per second, which is more than the combined speed of the next three most powerful supercomputers on earth, IBM said.

According to the Lawrence Livermore spokesperson David Schwoegler, ASCI White will do the heavy data crunching for the program, and the results will then be fed to SGI's clusters for visualization.