异构系统中带可用性约束的性能优化调度算法

摘要

针对可用性约束以及任务响应时间不同给异构系统中实时任务调度分配带来的处理机节点难选取的问题,提出了一种带可用性约束的性能优化调度算法(PO-SSAC).该算法将异构系统内处理机节点、实时任务以及带可用性约束的实时任务调度过程进行数学建模,通过考虑可用性成本与任务平均响应时间的折中,引入系统综合开销的概念,根据实时任务的可用性需求为其分配系统综合开销最少的处理机节点调度执行,达到系统调度资源合理利用的目的.实验结果表明,在实时任务调度可用性方面,与SSAC算法相比,PO-SSAC算法提升了3.4％,与MinMin算法相比,提升了76.9％,与Sufferage算法相比,提升了76.5％;与Sufferage算法相比,PO-SSAC算法的系统综合开销减少了约30％,优化了异构系统的实时任务调度性能.%A performance optimization scheduling strategy with availability constraints (PO-SSAC) for heterogeneous systems is proposed to address the processor choosing difficulty brought by availability constraints and different task response times during real-time tasks scheduling in heterogeneous systems.The strategy models processors,real-time tasks as well as the process of real-time task scheduling with availability constraints of a heterogeneous system,and a concept of system comprehensive cost is introduced by considering the compromise of availability costs and average response times of tasks.Then tasks are allocated to processors with the least system comprehensive cost according to specific availability request of real-time tasks,which reaches the goal of reasonable utilization of scheduling resources within the heterogeneous system.Experimental results show that the real-time tasks scheduling availability of the proposed PO-SSAC is improved by 3.4％,76.9％ and 76.5％ compared with the SSAC algorithm,MinMin and Sufferage algorithms,respectively,and that the proposed strategy reduces the system comprehensive cost by nearly 30％ compared with the Sufferage.These results show that the proposed strategy optimizes the performance of real-time tasks scheduling in the heterogeneous systems.