8:30 AM-9:10 AM
Chair: Robert G. Voigt, National Science Foundation
Clusters of powerful workstations or PCs connected by various network technologies have become a popular, inexpensive computing resource. Massively parallel computers are still considered to be the right tool for the biggest scientific and engineering computations. However, from an application standpoint, the difference between the two architectural options seem to be getting less and less obvious. The cluster approach can scale to very large configurations effectively, although perhaps not to equal the level of the very biggest MPPs. Perhaps the most important issues have to do with how the system is managed (as one powerful system or as many workstations), and on the entire system environment, including the type of job scheduling, system software, and software tools, and the availability of mass storage and archival file systems.
This talk will explore these issues and present data from a variety of applications and configurations in use today.
Paul C. Messina
CACR, California Institute of Technology
AN97 Homepage | Program Updates|
Registration | Hotel and Dormitory Information | Transportation | Program-at-a-Glance | Program Overview