GOVERNMENT

Cray today announced that it has won an order to deliver one of the world's most powerful supercomputers to the computing center for the U.S. Army Corps of Engineers. This order is part of a multi-order award with an aggregate value of over $23 million that Cray has received from the DoD HPC Modernization Program. In mid-2005, the Engineer Research and Development Center (ERDC) in Vicksburg, Mississippi, will begin operating a Cray XT3 supercomputer with over 4,000 processors and a peak performance of more than 21 trillion calculations a second (teraflops). The addition will more than triple the capability of the currently installed ERDC Major Shared Resource Center (MSRC) systems, which include a 64-processor Cray X1 vector supercomputer and a large Cray T3E supercomputer. The new system, one of the largest built by Cray, will secure the ERDC MSRC's position as one of the most capable high performance computing centers in the world. The ERDC supercomputing center is a vital link in the day-to-day chain that provides critical support to the nation's warfighters. The Center's capabilities are available to users around the world 24/7, and enable the Army and the Corps of Engineers to carry out complex numerical simulations that support both military and civil engineering missions. Although the supercomputing center is housed at ERDC, its focus is not only on the Army and Corps of Engineers missions. The MSRC supports DoD-wide research and development programs that utilize supercomputing capabilities for the Secretary of Defense, including "Challenge Projects," some of the most important computational projects in the Department. The new ERDC system will be the largest system in the Department of Defense's High Performance Computing Modernization Program (HPCMP) and one of the largest systems anywhere in the world. "This is a huge boost in computational capability for the Army and the Department of Defense," said Jeffery P. Holland, Director of the ERDC Information Technology Laboratory, which hosts the MSRC. "These systems form the computational backbone of the support we provide to the Defense mission. This new enhancement will enable us to continue to be a leader in providing computational capabilities and expertise for our DoD users worldwide." "The Department of Defense computational science community will need over 875 HABU equivalents of computing capacity by 2010. The Cray XT3 being installed at the ERDC MSRC, this year, is an important step toward meeting that requirement," said Cray Henry, Director of the DoD HPC Modernization Program (one HABU equivalent represents the equivalent performance of the weighted DoD benchmark suite on a 1024 CPU IBM power 3 system). "This system will support our most demanding computational research and engineering teams working on a variety of key technologies such as simulation of tactical aircraft performance, simulating high energy laser performance and simulation of armor performance. This system will provide DoD scientists and engineers with the leading computational capability they need." "Our users will be able to advance their critical research on one of the world's most powerful and efficient supercomputers," said John West, director of the ERDC Major Shared Resource Center. "The new Cray XT3 is a highly innovative supercomputer built to handle the most demanding scientific and engineering applications. We look forward to continuing our beneficial relationship with Cray." "Cray is excited that ERDC is making our new Cray XT3 technology available to their broad user base in support of ERDC's important national mission. The Cray XT3 supercomputer sets a new standard for the real-world performance, efficient scalability and reliability of system designs using standard microprocessors. We are fully committed to helping ERDC and the U.S. Department of Defense realize their ambitious goals for this new system," said Peter Ungaro, Cray senior vice president for sales, marketing and service. The Cray XT3 supercomputer uses AMD Opteron" processors connected via an innovative, high-speed internal network and is designed for very efficient, reliable operation at large scale. The system is based on the Red Storm architecture co-developed by Cray and Sandia National Laboratories.