APPLICATIONS

MOUNTAIN VIEW, Calif. -SGI (NYSE: SGI) today announced that a 512-processor SGI® Origin® 3800 supercomputer, installed in July at Los Alamos National Laboratory, N.M., is allowing scientists to complete intricate climate simulations in weeks rather than months. The installation of this 512-processor system is helping government and academic researchers across the country to better understand human activity and its relation to the earth's climate. By providing coherent shared memory across 512 processors, this SGI Origin 3800 supercomputer at Los Alamos National Laboratory has dramatically increased code performance and had a major impact on global climate modeling. Understanding the processes that control the earth's climate and predicting future changes in climate due to natural and anthropogenic influences require computer models of the climate system that describe the time-evolving circulation and thermodynamics of the atmosphere and oceans, the two main components of the climate system. The Parallel Ocean Program (POP) was developed at Los Alamos National Laboratory under the Department of Energy's Climate Change Prediction Program (formerly CHAMMP), which was established to rapidly advance the science of predicting the climate a decade or more in advance. Los Alamos National Laboratory has installed the POP model to run on the SGI Origin 3800 machine to perform high-resolution ocean model simulation. "Six years ago, we did a simulation of the North Atlantic climate over a 10-year period that took five months to complete. Now, with a more powerful, dedicated computer, we're substantially cutting the run time for these models. What used to take two months, now takes a week," said Robert Malone, a member of Los Alamos National Laboratory's climate ocean and sea ice modeling team. POP, which has become a standard model for many climate simulators, can now resolve one-tenth of a degree of the ocean's surface, the equivalent of 10 km at the equator. The finer resolution makes a difference because it enables the researchers to simulate ocean eddies, which play a key role in ocean dynamics. By utilizing the power of supercomputers, scientists can use numerical model simulation as an alternative method of studying various oceanographic and atmospheric phenomena, especially in the research of climate phenomena such as El Niño. Los Alamos National Laboratory's main objective is to optimize the parallel ocean model to run effectively on massively parallel processing supercomputers like the new 512-processor SGI Origin 3800 server. SGI Origin 3800, built on the SGI® NUMAflex™ shared-memory architecture, is the largest shared-memory system in the SGI® Origin® 3000 server series. The single-system image (SSI) consists of 512 processors attached to shared memory and an input/output subsystem, all of which are connected by the robust IRIX® operating system from SGI. The SSI system is unique in that it uses a single operating system to control its 512 processors, single shared memory and the input/output subsystem.