Silicon Graphics unveils plans for Intel based Teraflop/s ccNUMA System by Year 2000

Silicon Graphics will base the new systems on its MIPS microprocessor technology with straightforward, module swap upgrades to IA-64. These high-end servers will deliver Tflop/s compute performance and a Tbyte of memory in a single shared-memory system, enabling research and industry users to tackle problems and workloads only dreamed of until now.

"Phenomenal amounts of data continue to intensify the demands on high-performance systems. Uniting the scalable Silicon Graphics ccNUMA architecture with Intel IA-64 microprocessors will open up a whole new set of possibilities in the realm of supercomputing with unprecedented power and ease of use for attacking these problems," said Larry Smarr, director of the National Computational Science Alliance and the National Center for Supercomputing Applications. "Leading corporations are partnering with the Alliance and NCSA to learn how to use scalable computing to solve mission-critical data-intensive business analysis, as well as challenging technical problems. These corporations will find the Silicon Graphics Intel-based scalable systems an ideal enterprise-scale solution for these demanding computations."

"As the dominant provider of ccNUMA systems, with over 25,000 systems installed, we know how to crack the performance limits of a scalable system. Our second generation ccNUMA machines will have the highest bandwidth and lowest latency of any system available and will support MIPS and IA-64 microprocessors," said John R. "Beau" Vrolyk, senior vice president, Servers and Supercomputing Business Unit, Silicon Graphics. "We are driving what our customers really care about -- time to insight. The unique ability to drive the highest performance graphics and visualization directly from the memory of our ccNUMA systems allows our customers to solve problems, and understand the solution, in minutes rather than days. We have a customer who digitally prototyped 1,000 cars in the time it took to build one clay model. With our next generation ccNUMA machine it will be 2,000 to one. This is where pure performance gets down to work."