Robust Partitioning Policies of Multiprocessor Systems

Abstract Various techniques for multiprogramming parallel multiprocessor systems have been proposed recently as a way to improve performance. A natural approach is to divide the set of processing elements into independent partitions, and simultaneously execute a different parallel program in each partition. Several issues arise, including the determination of the optimal number of programs allowed to execute simultaneously (i.e., the number of partitions) and the corresponding partition sizes. This can be done statically, dynamically, or adaptively, depending on the system and workload characteristics. In this paper several adaptive partitioning policies are evaluated. Their behavior, as well as the behavior of static policies, is investigated using real parallel programs. The policy applicability to actual systems is addressed, and implementation results of the proposed policies on an iPSC/2 hypercube system are reported. The concept of robustness (i.e., the ability to perform well on a wide range of workload types over a wide range of arrival rates) is presented and quantified. Relative rankings of the policies are obtained, depending on the specific workload characteristics. A trade-off is shown between potential performance and the amount of knowledge of the workload characteristics required to select the best policy. A policy that performs best when such knowledge of workload parallelism and/or arrival rate is not available is proposed as the most robust of those analyzed.

[1]  John K. Ousterhout.  Scheduling Techniques for Concurrebt Systems. , 1982, ICDCS 1982.

[2]  John K. Ousterhout,et al.  Scheduling Techniques for Concurrent Systems , 1982, ICDCS.

[3]  Kenneth C. Sevcik.  Characterizations of parallelism in applications and their use in scheduling , 1989, SIGMETRICS '89.

[4]  John Zahorjan,et al.  Processor scheduling in shared memory multiprocessors , 1990, SIGMETRICS '90.

[5]  Lawrence W. Dowdy,et al.  Dynamic partitioning in a transputer environment , 1990, SIGMETRICS '90.

[6]  Shikharesh Majumdar,et al.  Scheduling in multiprogrammed parallel systems , 1988, SIGMETRICS '88.

[7]  Anoop Gupta,et al.  Process control and scheduling issues for multiprogrammed shared-memory multiprocessors , 1989, SOSP '89.

[8]  Lawrence W. Dowdy,et al.  On Modeling Partitioned Multiprocessor Systems , 1994, Int. J. High Speed Comput..

[9]  Satish K. Tripathi,et al.  An analysis of several processor partitioning policies for parallel computers , 1991 .

[10]  Mary K. Vernon,et al.  The performance of multiprogrammed multiprocessor scheduling algorithms , 1990, SIGMETRICS '90.

[11]  Shikharesh Majumdar,et al.  Characterisation of Programs for Scheduling in Multiprogrammed Parallel Systems , 1991, Perform. Evaluation.

[12]  Anoop Gupta,et al.  The impact of operating system scheduling policies and synchronization methods of performance of parallel applications , 1991, SIGMETRICS '91.

[13]  Tim Brecht,et al.  Processor-pool-based scheduling for large-scale NUMA multiprocessors , 1991, SIGMETRICS '91.

[14]  Leonard Kleinrock,et al.  Power and deterministic rules of thumb for probabilistic problems in computer communications , 1979 .

[15]  Satish K. Tripathi,et al.  The Processor Working Set and Its Use in Scheduling Multiprocessor Systems , 1991, IEEE Trans. Software Eng..

[16]  Lawrence W. Dowdy,et al.  Multiprogramming a Distributed-Memory Multiprocessor , 1989, Concurr. Pract. Exp..