Supercomputing Institute Research Bulletin Spring 1997

An Increasing Impact on High-Performance Computing at Minnesota-The University of Minnesota-IBM Shared Research Project In the two years since its inception in late 1994, the University of Minnesota-IBM Shared Research Project has become an integral part of the high-performance computing resources offered to the University of Minnesota research community. This project, which is jointly managed by the Computer Science Department, the Office of Information Technology, and the Supercomputing Institute, combines resources donated to the University by IBM as part of its Shared University Research (SUR) program, funding from the University, and additional support provided by the National Science Foundation CISE Infrastructure program. The University of Minnesota-IBM Shared Research Project provides state-of-the art technology for scientific computing combined with full-time, on-site user support and weekly tutorials. The project has been succesful in fostering a cooperative environment where researchers can interact with one another to fully realize the potential of this computer technology. During the past two years, forty-one University of Minnesota faculty principal investigators from eighteen departments have utilized these resources while working with 338 other researchers (primarily students and research associates) on sixty-eight peer reviewed research projects. These research projects are diverse in scope and include work by Geology Professor David Yuen on modeling thermal convection in the Earth’s mantle, and research by Computer Science Professor David Du aimed at developing computer video delivery systems (see sidebar on page four for a more detailed description of this research project). The high-performance computing hardware available through the University of Minnesota-IBM Shared Research Project offers great flexibility and computing power. It is possible to run a wide variety of high-performance computing jobs using the eight IBM RS/6000 workstations and the two frames of the IBM SP supercomputer. The system also has room for expandability and is steadily being upgraded. In April of 1997 the available configuration will be upgraded to a total of 24 processors. The IBM SP supercomputer will have four Wide nodes, each with one gigabyte of memory and between four and 12 gigabytes of disk storage, ten Thin2 nodes with 256 megabytes of memory and between four and six gigabytes of disk storage, and two Thin3 nodes with 512 megabytes of memory and nine gigabytes of disk storage. Four of the 590 workstations have 256 megabytes of memory, and four have 512 megabytes of memory. Each of the 590s has, at a minimum, two gigabytes of local scratch storage. Further details on the configuration are available on the World Wide Web: http://www.cs.umn.edu/Research/ibm-cluster/. The two forms of the IBM SP supercomputer, located in the Electrical Engineering/Computer Science Building on the Minneapolis Campus A unique component of this project is the Partnership Program that allows researchers to contribute nodes to the cluster. In exchange for the node the partner receives a significant allocation of resources which can be used on the entire cluster, receives access to cluster software, avoids frame costs for the SP nodes, and is not charged for the management of the contributed node. The benefits for the entire program include the additional resources that are available from the added node to the University’s research community, the advantage of having a larger cluster for large, path-breaking, jobs, and the nonlinear interactions that arise when the number of researchers utilizing this resource increases. The University of Minnesota-IBM Shared Research Project goes beyond the installation of state-of-the-art technology and its use in world-class research. It has also allowed for the development of a close and symbiotic relationship between the University and IBM. IBM staff members attend and contribute to the monthly meetings of the Computer Science-Supercomputing Institute Workstation-Office of Information Technology Cluster Steering Committee. This faculty committee has worked closely with IBM staff, in particular Pat Carey and Norman Troullier, on long-term planning. Furthermore, joint projects of University faculty and IBM have played a role in research and product development at IBM’s Rochester Minnesota facilities. A noteworthy event that grew out of this relationship was the International Conference on Parallel Computing which was held at the Institute October 3-4, 1996. The conference, which was sponsored by the Supercomputing Institute and IBM, brought together an internationally renowned group of speakers to discuss the progress and direction of parallel computing. There were seventeen plenary lectures and nine poster papers. Several of the talks focused on SP applications. Gyan Bhanot of IBM Thomas J. Watson Research Center in Yorktown Heights, New York, described SP applications in nine different areas. (A complete article on the conference is available in the Winter 1996-97 Supercomputing Institute Research Bulletin and is accessible via our web site: http://www2.msi.umn.edu/Bulletin/Vol.13-No.2/Winter1997.html. Four of the eight IBM RS/6000 workstations available through the University of Minnesota-IBM Shared Research Project Research results from this project are listed in each Supercomputing Institute Research Bulletin in the Research Reports section. Through March 1997, researchers have submitted thirty-nine research reports to the UM-IBM research report series. These reports have both a UMSI and a UM-IBM report number. A complete list of UM-IBM publications is available on the World Wide Web: http://www.cs.umn.edu/Research/ibm-cluster/reports.html.

In This Issue: Cray T3E Upgrade > University of Minnesota-IBM Shared Research Project Design and Development of Multimmedia Servers High Performance Computing in Geography Research Solvation Model Research Reports

 

This information is available in alternative formats upon request by individuals with disabilities. Please send email to alt-format@msi.umn.edu or call 612-624-0528.
 

	URL: http://
This page last modified on
Website related questions or problems should be directed to webmaster@msi.umn.edu The Supercomputing Institute does not collect personal information on visitors to our website. For the University of Minnesota policy, see www.privacy.umn.edu.