Improved Master-Slave Parallel Task-Allocating Algorithm and Its Implement

摘要

This paper is based on the research of master-slave parallel task allocation, and points out the algorithm not considering the cost of the task which will be added to processor. Considering that would cause the disequilibrium of task allocation on slaving processors, this paper presents the improved master-slave parallel task-allocating algorithm. Before task groups being assigned to slave-processors, the main processor makes all the task groups being in descending order of their cost. Avoiding costly task groups concentrating in the busy processor, this manipulation makes the free processor getting the most costly task group of the rest task groups at all time and achieves the goal of load balance. The experimental results show the advantages of the new algorithm and certify that it efficiently solves the bad balance problems, and improves the efficiency.