CLOUD

By Steve Fisher, Editor In Chief -- On February 4th, RackSaver and its partners Dolphin ICS, Scali and AMD announced the delivery of a new clustered supercomputer system to the University of California Santa Cruz (UCSC). The system will primarily be used in planetary physics research and consists of 132 Linux-based nodes configured with 264 AMD Athlon MP processors, 132 gigabytes of memory and nearly 8 terabytes of storage. Referred to as UpsAnd after the Upsilon Andromedae Planetary System, the system’s peak performance is reportedly in excess of 300 gigaflops. To learn more, Supercomputing Online sat down with Paul Mecucci, Racksaver’s Sales & Marketing Manager for the following brief interview. For additional background click RackSaver Builds a 264 AMD Athlon MP Processor Supercomputer for UC Santa Cruz. SCO: Please provide the readers with some technical specifics on the cluster recently delivered to UC Santa Cruz. MECUCCI: The Cluster was built with 132 Dual AMD AthlonMP systems. 264 AMD 1.4GHz Processors total. Each Node (referring to one system) contains 1GB of DDR memory. The High Speed interconnect is the Dolphin SCI, which allows for high speed communications between each node. This allows the individual systems to act as one powerful supercomputer. SCO: If you can, please give us some details on how UCSC plans to use the cluster, citing specific applications and areas of special interest to researchers there. MECUCCI: Well, I'm afraid I can only supply a very basic description of how the UCSC scientists will use UpsAnd and it is this...they'll use it to learn more about planets, stars, and the universe by modeling them. As far as special interests, I know that Gary Glatzmaier, the principal investigator at UCSC is into three-dimensional simulations of the convection and magnetic field generation in the fluid interiors of planets. He produced the first self-consistent simulation of the Earth’s magnetic field. I believe some other plans are using the system for modeling the evolution of the early universe and simulating supernovae explosions. SCO: Are clustered systems such as UpsAnd particularly well-suited to applications such as planetary physics? MECUCCI: No, our clusters can be used in almost any market, from Oil & Gas research, Genetic research, Video rendering...The list goes on and on. The software would be the determining factor as to how the system will be specifically used. What stops a lot of research entities from going to an X86 scalable Linux cluster is software. SCO: How many systems such as UpsAnd has RackSaver delivered and to whom? Primarily universities and other research institutions? Any corporate customers? If so how are they exploiting RackSaver technology? MECUCCI: We have delivered over 5,000 nodes to Vertas DGC. They use the cluster for Seismic Data processing. They have the nodes divided into 64 nodes clusters. Our technology allows for up to 160 processor sin a single 7' cabinet. This is the most processing power that can be stacked in a single cabinet. The cluster is made up of our RS-1200 chassis. SCO: What's on the horizon for RackSaver as a company in terms of new products, corporate growth, alliances, etc.? MECUCCI: We are constantly working with the latest processors from the two largest processor vendors (AMD & Intel). This allows us to provide end users with the newest technology packed in a slimline Rack-optimized Server. We have a huge product release coming in about a month or so that I am not privileged to speak about yet. We will have a large release on that soon.