SiCortex, the first company to engineer a Linux cluster from the silicon up, today announced that the first production model of an SC5832, its flagship 5.8 teraflop system, will be installed at the U.S. Department of Energy's Argonne National Laboratory in Argonne, Illinois. The lab and its community of researchers will take advantage of the unique capabilities and energy efficiencies of the SC5832 to conduct research in a variety of areas, including astrophysics, climate modeling, oil and gas exploration, seismic research and biotechnology.
The SC5832 is a high performance computer that dramatically reduces power usage while providing industry leading performance. A SiCortex cluster node consumes 15 watts of power, an order of magnitude less than the 250 watts used in a conventional cluster node.
As we move into the era of petascale computing, scaling current applications to work with thousands of processors will be a major challenge, said Rick Stevens, associate laboratory director of Computing and Life Sciences at Argonne. We believe that the power-efficient SiCortex architecture represents the way high-performance computers will be designed in the future.
SiCortex has introduced a new concept in high-performance computing by implementing a complete cluster node on a chip, including six 64-bit processors, multiple memory controllers, a high-performance cluster interconnect and a PCIexpress connection to storage and internetworking. The SC5832 can perform six trillion operations per second in a cabinet that is less than one-third the size of conventional clusters.
There is no more competent and motivated team than at Argonne, said SiCortex CEO Dr. John Mucci. Their interests in existing applications, in fostering innovative new applications, and in exploring and utilizing green, energy-efficient new computer architectures make Argonne an ideal partner. In particular, we share their belief that open source software is a powerful paradigm for moving applications forward.
High-performance computers increasingly consist of many thousands of processors, presenting unique software challenges. In order to operate effectively, these large-scale computers require extremely fast communications between processors and substantial I/O bandwidth. The SC5832 boasts the fastest communications and I/O of any computer in its class running current applications.
Argonnes Mathematics and Computer Science (MCS) Division has a long tradition of acquiring, evaluating and deploying advanced high-performance computing architectures. Moreover, the division has been a leader in developing open source software that is widely used throughout the HPC community.
The SiCortex platform is completely open source, from the operating system to the job scheduler, said Ewing Lusk, director of Argonnes MCS Division. In addition to solving challenging computational problems, the machine will benefit the whole computer science community, as they improve, extend, and contribute to the open source software community for petascale platforms.
|Contact: Eleanor Taylor|
DOE/Argonne National Laboratory