基于多核处理器平台的公平调度算法

Fair scheduling algorithm on multi-core platforms based platforms

为了减少多核处理器系统全局调度算法中共享L2cache抖动,在Pfair调度算法基础上提出一种新的Cache感知的软实时公平调度算法.通过对WSS(work set size)、子任务截止时间和任务负载建立多因素优先级模型,并将此优先级模型应用到改进后的Pfair算法中,该算法的调度决策在系统负载较重的系统中对WSS具有自适应性.模拟实验结果显示:在对称4核和8核处理器系统中,该算法任务丢失率低,且在系统负载重时能够减少共享L2cache抖动,其整体调度性能优于传统的G-EDF(global-earliest deadline first)调度算法和Pfair调度算法.

A new cache-aware scheduling algorithm based on Pfair scheduling algorithm for soft real-time tasks on multi-core platforms was proposed to reduce the shared L2cache thrashing of global scheduling algorithm on multi-core processors.The task priority model of the proposed algorithm which was used in the extended Pfair scheduling algorithm was established by the work set size（WSS）,the sub-task deadline,and the task utilization.The scheduling decision of the proposed algorithm is self-adaptive for WSS on heavy-loaded systems.Comparing with the traditional global-earliest deadline first（G-EDF） scheduling algorithm and the Pfair scheduling algorithm,the proposed algorithm can obtain lower task deadline miss rate and the L2-cache thrashing rate on the heavy-loaded multi-core processors.Consequently,the overall performance of the proposed algorithm is better than the other two scheduling algorithms.