NOAA supercomputer to boost hurricane forecasts

NOAA supercomputer to boost hurricane forecasts

$34 million contract calls for two SGI Origin 3800 clusters to perform simulations and analyses

BY PATRICIA DAUKANTAS | GCN STAFF

National Oceanic and Atmospheric Administration researchers expect their newest high-performance computer to advance the science of hurricane forecasting.

NOAA recently awarded a $34 million contract to Raytheon Co. to configure the supercomputer at the agency's Geophysical Fluid Dynamics Laboratory in Princeton, N.J. It will consist of two SGI Origin 3800 clusters, said Bruce Ross, the laboratory's deputy director. The Origin 3800 is SGI's latest high-performance system [GCN, Aug. 7, 2000, Page 8].

One eight-node cluster, which NOAA officials call the large-scale cluster, will run huge fluid dynamics simulations of atmospheric and oceanic behavior. Each node will have 128 400-MHz Mips R12000 processors, Ross said.

The other cluster, consisting of two 64-processor nodes, will perform statistical analysis and visualization routines on the results, Ross said. The smaller cluster also will serve as a front end to the hierarchical storage management system that will organize the data sets of weather models.

The storage system will consist of three robotic tape silos from Storage Technology Corp. of Louisville, Colo. Initially the archive will hold 500T of data, but Ross said capacity will increase to 2 petabytes, or 2,000T, by September 2003.

The system's theoretical peak speed will be roughly 900 billion floating-point operations per second.

NOAA is leasing the large-scale cluster and a portion of the analysis cluster from Raytheon's Command, Control, Communications and Intelligence Division in Garland, Texas. The agency is acquiring the remaining hardware through a lease-to-purchase arrangement, Ross said.

The base contract with Raytheon runs through the end of fiscal 2003. According to the contract schedule, the system will be upgraded to 10 nodes, each with 128 700-MHz R16000 processors, in June 2002.

inside gcn

  • artificial intelligence (ktsdesign/Shutterstock.com)

    Machine learning with limited data

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

More from 1105 Public Sector Media Group