"This kind of recognition spotlights the computational power attainable with a Beowulf cluster system built with the Dolphin Interconnect WulfKit, the high-speed, low-latency scalable coherent interface (SCI) interconnect", stated Keith Murphy, Dolphin VP of Sales. "One of the advantages is the substantially lower cost per unit of computing power of a WulfKit cluster as compared with a mainframe of comparable computing capacity and other clusters using slower interconnects", Mr. Murphy explained.
The TOP500 Supercomputers list is updated twice annually and published by Mannheim University and has been published since June 1993. Supercomputers are ranked on the TOP500 according to performance levels measured by the LINPACK Benchmark while solving a dense system of linear equations.
The Umea University supercluster is installed at the High-Performance Computing Center North (HPC2N). The 120 nodes in the supercluster, also known as a Beowulf cluster, are interconnected as a three-dimensional 4x5x6 "torus," or logical cube. The superior performance, which earned the supercluster the 94th position on the Top 500 Supercomputer list, is achieved with 240 AMD Athlon? MP2000+ processors in 120 dual nodes and interconnected with the innovative Dolphin WulfKit3, a high-speed interconnect that features a three-dimensional torus distributed switch topology.
The University of California Santa Cruz supercluster is a 132-node, 264-processor system featuring AMD Athlon MP processors. The system was installed earlier this year at the Center for the Origin, Dynamics, and Evolution of Planets (CODEP) located within the Institute for Geophysics and Planetary Physics at the University. The supercluster supports a wide range of geophysics and astrophysics research projects and is also used to train undergraduate and graduate students in parallel programming techniques.
The STAAG Imaging 64-node 128-processor supercluster also features AMD Athlon MP processors. STAAG Imaging provides time and depth imaging services to the oil and gas industry where quality and turnaround are critical issues, both of which are supported at unprecedented levels in this industry niche by the supercluster system.
The Dolphin WulfKit system is a standards-based high-speed interconnect used to cluster large numbers of servers creating a supercluster with performance levels comparable to mainframe supercomputers. These WulfKit-enabled superclusters are typically used in highly demanding parallel processing applications in business and research.
WulfKit uses the IEEE 1496 scalable coherent interface (SCI) introduced about ten years ago as a memory-to-memory interconnect standard. A key feature of SCI is its very high speed and low latency, which the WulfKit interconnect system maximises. Dolphin Interconnect Solutions provides a family of high-speed interconnects which exploit several key features of the SCI standard. WulfKit is a combination of a PCI SCI Adapter Card developed by Dolphin Interconnect, and a powerful message-passing interface (MPI) implementation developed by Scali in Oslo, Norway.
Previously, a Dolphin/WulfKit supercluster system installed at the Center for Parallel Processing in Paderborn, Germany first appeared on the 13th TOP500 list published in June 1999.