网络计算环境下任务调度问题研究

利用网上空闲处理机组成机群并行计算环境是快速增长的研究领域,任务调度是其中最关键性的问题之一.在基于消息传递机制的网络并行计算环境下,针对一类大粒度计算任务并行时协同同步工作问题,发展了一种新的静态调度模型,提出了调度算法,它映射一类特殊的并行程序任务到由网络上若干空闲处理机组成的机群中.研究表明该算法能减少此类并行程序的完成时间.

Research of Task Scheduling in NetworkParallel Computing Environment

Clusters of workstations on common network are much increasing being used as cost effective parallel computing platforms. Task scheduling is one of the crucial problems in cluster computing. A new scheduling graph model is developed to present a kind of special parallel program, which solves synchronous working problem among coarse grained tasks in message passing system. A scheduling algorithm that maps this kind of program onto a cluster is provided, which consists of some idle workstations in a LAN for parallel computing. It is demonstrated that algorithm can reduce the completion time of the parallel program.