Non-contiguous processor allocation algorithms for distributed memory multicomputers
- Univ. of Oregon, Eugene, OR (United States). Dept. of Computer and Information Science
- NASA Ames Research Center, Moffett Field, CA (United States). NAS Systems Division
Current processor allocation techniques for highly parallel systems have thus far been restricted to contiguous allocation strategies for which performance suffers significantly due to the inherent problem of fragmentation. The authors are investigating processor allocation algorithms which lip the restriction on contiguity of processors in order to address the problem of Fragmentation. Three non-contiguous processor allocation strategies: Naive, Random and the Multiple Buddy Strategy (MBS) are proposed and studied in this paper. Simulations compare the performance of the non-contiguous strategies with that of several well-known contiguous algorithms. They show that non-contiguous allocation algorithms perform better overall than the contiguous ones, even when message-passing contention is considered. They also present the results of experiments on an Intel Paragon XP/S-15 with 208 nodes that show non-contiguous allocation is feasible with current technologies.
- OSTI ID:
- 87626
- Report Number(s):
- CONF-941118-; ISBN 0-8186-6605-6; TRN: IM9535%%258
- Resource Relation:
- Conference: Supercomputing `94 meeting, Washington, DC (United States), 14-18 Nov 1994; Other Information: PBD: 1994; Related Information: Is Part Of Supercomputing `94: Proceedings; PB: 849 p.
- Country of Publication:
- United States
- Language:
- English
Similar Records
Processor allocation in an N-cube multiprocessor using Gray codes
Multigrid and Krylov Solvers for Large Scale Finite Element Groundwater Flow Simulations on Distributed Memory Parallel Platforms