RETAIL

As scientists mark the 50th anniversary of operational numerical weather prediction, Silicon Graphics announced that its high-performance computing technology continues to help drive innovations in weather forecasting and climate research the world over. SGI technology powers weather forecasting in half of the world's weather centers, while high-end SGI systems enable researchers to speed the search for answers to complex questions on environmental trends and global warming. Environmental scientists and researchers at a broad range of facilities -- including meteorological and oceanographic centers, ground stations, U.S. national laboratories and NASA -- are increasingly turning to SGI servers and storage solutions to predict short-term weather developments or identify long-term patterns and cycles. As a top-performing platform for leading numerical weather prediction applications and complex modeling solutions, scalable SGI systems are also prompting a migration from proprietary UNIX(R) OS-based to cost-efficient industry-standard technologies. "Weather prediction and research are crucial to the daily operations of industries and governments across the globe," said Dr. Jill Matzke, manager of environment marketing at SGI. "As environmental scientists strive to learn more by collecting and analyzing ever larger data sets and models, their computational and storage requirements increase significantly. SGI has such a broad presence in this market because we are committed to providing solutions that give these customers the scalability, massive memory bandwidth and high-speed data management solutions they need to store and process vast amounts of data." Unparalleled Access to Time-Sensitive Data SGI solutions are in widespread use in several key areas of weather prediction and research: Observational acquisition and processing; numerical weather prediction (NWP), climate research and environmental modeling; and data management throughout the research and operational forecast workflow. Keys to SGI's broad acceptance include dramatic and reliable system scalability -- with linear performance scalability to 512 processors and more on SGI servers -- and industry-leading memory bandwidth. In 2003, for instance, an SGI system became the first supercomputer in history to break the 1 terabyte/second barrier on the STREAM Triad benchmark, an industry standard measurement of memory bandwidth. These technology advantages provide an edge for environments like operational forecasting, where productivity factors such as time-to-solution are paramount. "To produce more accurate forecasts, we must continually increase the amount of satellite data, imagery and other information we analyze while also amplifying the resolution of our models," said Alain Beuraud, Head of HPC and Storage Department at Meteo-France, the leading source of weather information and forecasts in France. "This produces a huge amount of data that must be stored, managed, accessed and analyzed with extraordinary efficiency if we are to predict tomorrow's weather in time to make a difference. SGI solutions give us the bandwidth, scalability and real-time access to data that this type of weather prediction demands." The high-bandwidth architecture and reliable scalability of SGI systems also has driven a host of performance breakthroughs in environmental sciences. For instance, SGI offers one of the industry's fastest platforms for MM5, the widely used numerical weather prediction model from Pennsylvania State University and the National Center for Atmospheric Research (NCAR). SGI technology also powers one of the fastest platforms for the WRF (Weather Research and Forecast) Modeling System, a numerical forecast system being developed and managed by the Department of Commerce (NOAA), the Department of Defense (Air Force and Navy), the Department of Transportation (FAA), and the University Corporation for Atmospheric Research (UCAR/NCAR). Similar performance advantages have led to broad adoption of SGI solutions at some of the world's premiere weather forecast and research facilities, including the U.S. Navy's Fleet Numerical Meteorology and Oceanography Center. Fleet Numerical operates 24 hours a day, 365 days per year and distributes meteorological and oceanographic information to military users around the world based on sophisticated global and regional numerical models. For instance, Fleet Numerical uses its Coupled Ocean/Atmosphere Mesoscale Prediction System, COAMPS(R) model to predict weather and ocean conditions in highly complex coastal areas of the world such as the Persian Gulf. SGI Driving Migration to Open Standards-Based Solutions SGI also is fueling the migration from proprietary UNIX systems to cost-effective, open standards-based solutions for prediction and research. At NASA Ames Research Center, for example, an SGI server powered by a 64-bit Linux(R) operating environment and 512 Intel(R) Itanium(R) 2 processors has helped NASA researchers achieve the fastest result ever recorded for the ECCO (Estimating the Circulation and Climate of the Ocean) application. The supercomputing-class system also represents the first 512-processor system to operate under a single instance of the Linux kernel. At the National Leadership Computing Facility at the U.S. Department of Energy's Oak Ridge National Laboratory (ORNL), researchers have deployed industry-standard SGI servers and storage solutions to conduct sophisticated climate modeling in an effort to project the potential long-term impact of such environmental threats as pollution and ozone depletion. The ORNL installation leverages key SGI data management advantages, including access to data via Storage Area Networks (SANs) and high-availability RAID arrays. SGI storage solutions prove ideal for data-intensive weather research environments that require storage and archive solutions that can grow easily, cost-effectively and continuously without limitations. "Data management is a core aspect of our studies in global warming and climate trends," said Thomas Zacharia, associate lab director for ORNL. "Research is only as good as the information used to conduct it, and efficiently accessing and using that information lies at the heart of a successful climate research effort. SGI technology has proven enormously helpful in this regard, both for processing information in a shared-memory environment and for storing and retrieving it across a SAN." For more information about SGI industry solutions visit http://www.sgi.com/industries/ .