‘Falcon’ adds power to support mission-critical work Published July 5, 2006 By the Advanced Computational Analysis Directorate WRIGHT-PATTERSON AIR FORCE BASE, Ohio -- The newest supercomputer at the Aeronautical Systems Center’s Major Shared Resource Center went into production recently, enabling advanced problem-solving capabilities for mission-critical support to the Department of Defense. The new Falcon system will aid the ASC MSRC in providing greater computer resources to the DoD research and development and test and evaluation communities. ASC’s MSRC is one of four such High-Performance Computing centers throughout DoD. The Falcon system consists of 2,048 processors or 1,024 XC nodes with 2 (2.8 GHz) AMD Opteron central processing units. Falcon expands the MSRC’s capability to more than 6,900 processors spread across 11 systems. TOP500 list ranking for this system is anticipated to be announced following this week’s (June 27 – 30) International Supercomputer Conference in Dresden, Germany, at www.top500.org. After system delivery, the ASC MSRC staff worked closely with Hewlett-Packard, the developers of the Falcon, to prepare the machine for the first phase of Effectiveness Level Testing, or ELT1. This period tests the reliability of the system to assure a smooth transition into the ASC MSRC HPC environment. After successful completion of ELT1 in mid-May, approximately 30 “Pioneer” users were given time on the system for the second phase of Effectiveness Level Testing, or ELT2. This phase helped to integrate the Falcon fully into the ASC MSRC HPC environment and address compatibility and configuration issues. The Falcon Pioneer program completed with full acceptance of the system on June 15. The Falcon went into full production June 21. “Several of our Pioneer users had positive comments about their experiences while running jobs on the Falcon,” said Steve Wourms, director of the ASC MSRC. “Pioneer users reported their delight with the Falcon as a well-balanced, high-speed system that will become a major ‘work horse’ for them with good input/output, good communications, and good compute power.” The HP Cluster Portfolio, XC System Software and Scalable File Share are part of HP’s Unified Cluster Portfolio, which helps customers manage and visualize large volumes of data. The 11.5 TeraFLOPS system has a high-speed Voltaire InfiniBand interconnect that is used with HP’s XC system software providing cluster-management capability. “This is the first large-scale Linux cluster that the ASC MSRC has ever integrated,” Mr. Wourms said. “We look forward to this system helping to accelerate research that will provide mission-critical support for our user community.”