CLOUD

SANTA CRUZ, CA -- RackSaver(TM) and its partners are helping to push the knowledge envelope further with a new supercomputer cluster for researchers in the fields of planetary physics and astrophysics at the University of California, Santa Cruz (UCSC). The supercomputer, with processing power in excess of 300 gigaflops, has been dubbed "UpsAnd" for the Upsilon Andromedae Planetary System, a star orbited by three planets that provided the first evidence that other solar systems like our own exist elsewhere in the universe. "UpsAnd" will be used to help UCSC scientists learn more about planets, stars and the universe by modeling these systems on their new supercomputer. The Supercomputer consists of 132 Linux based nodes configured with 264 AMD Athlon(TM) MP processors, 132 gigabytes of memory and nearly 8 terabytes of storage. Each node is connected with Dolphin Interconnect's high-performance WulfKit, which includes a 64Bit/66MHz SCI card and Scali high-performance MPI and Universe cluster management software. Only the RackSaver RS-1132 chassis allows for these specifications to be configured in a 19" 1U rack-optimized server design. With performance numbers exceeding 300 gigaflops, "UpsAnd" ranks among the top 100 supercomputers in the world, and currently is the most powerful high-performance supercluster San Diego-based RackSaver has built to date. "'UpsAnd' is the best example of the success we've had building RackSaver supercomputer clusters integrated with the Dolphin Wulfkit," said David Driggers, RackSaver(TM) CEO. "UpsAnd" is also the most powerful cluster to date using the Dolphin Wulfkit cluster interconnect. Despite its substantial computing capacity, the system is contained in just four standard server cabinets occupying about 24 square feet of floor space, a major breakthrough for density and performance. "UpsAnd" was specifically designed to specifications provided by the UCSC researchers who will be using the supercomputer. Principal investigator Gary Glatzmaier, a professor of the Earth sciences at UCSC, computes three-dimensional simulations of the convection and magnetic field generation in the fluid interiors of the planets and starts. He produced the first self-consistent simulation of the Earth's magnetic field, including spontaneous dipole reversals. "UpsAnd" will also be used to compute three-dimensional models to study asteroid and planetary collisions. Other planned projects include simulating supernovae explosions, modeling the evolution of large-scale structure in the early universe, and studying electromagnetic sound waves through turbulence in the atmosphere and oceans. "Critical, compute-intensive enterprise applications such as those being designed at UC Santa Cruz require the reliability and stability that an AMD Athlon(TM) MP processor-based system can provide," said Ed Ellett, vice president of Marketing at AMD. "UC Santa Cruz needed a supercomputer cluster that was configured to its specifications, and through RackSaver(TM) was able to get the performance they needed in a high-density, space-saving environment designed with AMD Athlon MP processors." "UpsAnd" can accomplish such high-level computations in part because its architecture differs from other cluster computers. In a typical cluster, the processors perform parallel tasks to compress the amount of time needed to process data. "UpsAnd's" 264 processors actually work together to complete a single task, similar to the behavior of a true supercomputer. "UpsAnd" is eight times more powerful than its "younger brother," a 32-node Beowulf cluster delivered to UCSC in March 2001 by RackSaver Inc.