Cluster Computing:

DELL CLUSTER AT NASA'S JPL: ONE OF TOP SUPERCOMPUTERS

A recently deployed Dell high-performance computing cluster (HPCC) at the Jet Propulsion Laboratory (JPL) in California is ranked as one of the 50 fastest supercomputers according to the Top 500 Supercomputing List. The ranking was announced Monday at the SC2004 conference on high-performance computing, networking and storage.

JPL, managed for NASA by the California Institute of Technology, has used "big iron" supercomputers for its research and engineering projects since the late 1980s. However, Dr. Chris Catherasoo, supervisor of JPL's Supercomputing and Virtualization Facility, and his colleagues noticed that clusters of standards-based server systems increasingly ranked among the top supercomputing sites.

"More and more of the top 500 supercomputers are cluster-based," Catherasoo said. "As 'big iron' isn't as cost-effective as standards-based clusters, we decided it was time to move to a solution with better price for performance."

The Dell cluster, to be used for earth and space science research, as well as space craft engineering and design, is comprised of 512 Dell PowerEdge 1750 servers with dual 3.2 GHz Intel Xeon processors. The cluster, running Red Hat Linux, has performed up to 4.3 trillion floating-point operations per second (TFLOPS) on the High-Performance Linpack (HPL benchmark).

About the JPL Supercomputing and Visualization Facility

The JPL Supercomputing and Visualization Facility, formed in May 2000, succeeded the JPL Supercomputing Project established in 1989. Its mission is to provide all JPL engineers and scientists with the best possible state-of- the-art high-performance computing and visualization environment to help them solve their computationally-intensive problems. Visit http://sc.jpl.nasa.gov for more information.