A Simple Yet E ective Load Balancing Extension to thePVM Software
نویسنده
چکیده
In this paper, a load balancing extension for the PVM software system is described. This extension includes slave processes, for obtaining host CPU load information, and a master process for gathering the load information. A programming interface is included to provide the PVM user with CPU load information and a load balancing process spawn. For tested applications , the load balancing extension exhibits considerable improvement in parallel runtime performance. Results presented include parallel runtime comparisons for basic matrix oriented algorithms, a parallel fractal image compression algorithm, and selected NAS parallel benchmarks.
منابع مشابه
An Effective Load Balancing Policy for Geometric-Decaying Algorithms
Parallel algorithms are often rst designed as a sequence of rounds where each round includes any number of independent constant time operations This so called work time presentation is then followed by a processor scheduling implementation on a more concrete computational model Many parallel algorithms are geometric decaying in the sense that the sequence of work loads is upper bounded by a dec...
متن کاملCost - E ective Data - Parallel Load Balancing
Load balancing algorithms improve a program's performance on unbalanced datasets, but can degrade performance on balanced datasets, because unnecessary load redistributions occur. This paper presents a cost-e ective data-parallel load balancing algorithm which performs load redistributions only when the possible savings outweigh the redistribution costs. Experiments with a data-parallel polygon...
متن کاملDistributed Interaction with Computon
One of the major challenges of distributed arti cial intelligence is to obtain useful and e ective emergent behaviors of agents in the system based on the local decision{making of each agent. The e ectiveness of the system as a whole is as much dependent on the form of interactions between agents as on the capabilities or strategies of each one. The focus and main theme of this paper is to put ...
متن کاملMIYAZAKI and YOKOTA : PARALLEL PRODUCTION SYSTEMS USING SOFTWARE CACHE
Because the match phase in OPS5-type production systems requires most of the system's execution time and memory accesses, we proposed hash-based parallel production systems, CPPS (Clustered Parallel Production Systems), based on the RETE algorithm for distributed memory parallel computers, or multicomputers to reduce such a bottleneck. CPPS was e ective in speeding up the match phase, but still...
متن کاملA Steady State Analysis of Di racting Trees
Di racting trees are an e ective and highly scalable distributed pa rallel technique for shared counting and load balancing This paper presents the rst steady state combinatorial model and analysis for di racting trees and uses it to answer several critical algorithmic design questions Our model is simple and su ciently high level to overcome many im plementation speci c details and yet as we w...
متن کامل