Saturday, March 15

1:30 PM-3:30 PM
Greenway I-J

Resource Scheduling/Management for Parallel Scientific Computing

Parallel processing systems make it possible to solve large, complex scientific problems that are otherwise intractable or prohibitively expensive. The scheduling policies that allocate/manage system resources among the applications submitted for execution have a significant impact on the performance of these scientific computing environments. Software support and algorithms are needed at both the system and application levels to achieve the best performance and to ease the management of system resources. The full benefits of parallel processing for scientific computing can only be realized by exploiting optimal, or near-optimal, resource scheduling methods at both the system and application levels.

The speakers in this minisymposium will present new scheduling methods and systems and novel scheduling approaches and systems that are shown to provide significant performance improvements over existing methods.

Organizer: Mark S. Squillante
IBM T. J. Watson Research Center

1:30 Modeling the Cost of Redistribution in Scheduling
Gary Shao, Rich Wolski, and Francine Berman, University of California, San Diego
2:00 Scheduling in a High Performance Remote-Sensing Data Server
Chialin Chang, Alan Sussman, and Joel Saltz, University of Maryland, College Park
2:30 Extensible Resource Scheduling for Parallel Scientific Applications
Nayeem Islam, IBM T. J. Watson Research Center, Andreas Prodromidis, Columbia University, Mark S. Squillante, Organizer, and Ajei Gopal and Liana Fong, IBM T. J. Watson Research Center
3:30 Optimal Scheduling of Coarse-Grained Parallel Scientific Applications
Mark S. Squillante, Organizer and Konstantinos Tsoukatos, University of Maryland, College Park

PP97 Homepage | Program Updates|
Registration | Hotel Information | Transportation | Program Overview | Program-at-a-Glance

MMD, 1/24/97