Media: Mardi Larson, 612/683-3538 Financial: Bill Gacki, 612/683-7372 NCAR ORDERS FOUR CRAY RESEARCH SUPERCOMPUTER SYSTEMS INCLUDING A CRAY T3D MASSIVELY PARALLEL PROCESSING SYSTEM EAGAN, Minn., April 26, 1994 -- The Boulder, Colo.-based National Center for Atmospheric Research (NCAR) has ordered four Cray Research high performance computing systems, including a 64-processor CRAY T3D massively parallel processing (MPP) system, a CRAY Y-MP8I supercomputer, and two CRAY EL92 deskside supercomputing systems. Three of the systems are being installed at NCAR's Scientific Computing Division in Boulder. The fourth, a deskside CRAY EL92 system, has been installed at NCAR's Mesoscale and Microscale Modeling Division, also in Boulder. This is the first supercomputer installed at NCAR for dedicated use by a division outside of its Scientific Computing Division. All four systems were ordered in the first quarter of 1994. The two CRAY EL92 systems were installed in first quarter 1994. The CRAY Y-MP8I system and the CRAY T3D MPP system will be installed this summer. The new CRAY T3D MPP system will be tightly coupled with the CRAY Y-MP8I supercomputer to provide NCAR users with a computing environment that combines traditional parallel vector processing and MPP capabilities. The CRAY Y-MP8I has eight processors and 64 million words (megawords) of central memory and is an upgrade to an existing CRAY Y-MP supercomputer system at NCAR. Massively parallel processing systems are well suited for environmental modeling and weather forecasting applications because they can readily be broken down into small parts that can be spread across the many processors of an MPP system. According to NCAR officials, the CRAY T3D and CRAY Y-MP8I systems will provide the foundation for the establishment of the Climate Simulation Laboratory (CSL) at NCAR. The aim of the CSL will be to partner with other organizations in research initiatives of mutual interest in the areas of global and regional climate change. Strong university, government and industry participation is planned for the CSL, according to NCAR. "The establishment of the CSL with the new Cray Research systems will provide the appropriate range of resources to the researchers in our scientific community," said Dr. Bill Buzbee, director of NCAR's Scientific Computing Division. "With this heterogeneous computing environment, any given computing problem can be broken down into parts that can be run on the most appropriate system to achieve the highest possible speed, resolution and accuracy." Scientists in NCAR's Mesoscale and Microscale Modeling Division will use the CRAY EL92 to develop and enhance regional weather forecasting models. "The CRAY EL92 is an affordable system that gives us the ability to run models that previously took larger, more expensive supercomputers," said Joe Klemp, head, Mesoscale Prediction System. "We will continue to heavily utilize all the Cray Research systems operated by the Scientific Computing Division, but the CRAY EL92 gives us a dedicated computing resource for modeling mesoscale processes." "In 1977, NCAR was the first organization to acquire a Cray Research supercomputer," said John Carlson, Cray Research chairman and chief executive officer. "NCAR's continued selection of Cray Research computers demonstrates our success in meeting the changing needs of the environmental research and weather forecasting community with a full range of high performance computing products, including our MPP and parallel vector systems, and our departmental supercomputers." Carlson said the company has received 19 orders for the new CRAY T3D MPP system, which was unveiled in September 1993. NCAR is a national research center located in Boulder, Colorado. It is operated by the University Corporation for Atmospheric Research (UCAR), a consortium of 61 North American universities with doctoral programs in the atmospheric and related sciences, and is sponsored by the National Science Foundation. Cray Research provides the leading supercomputing tools and services to help solve customers' most challenging problems. ###