基于多核处理器的关联任务并行感知调度算法

关联任务在多核处理器上并行调度所产生的通信时延，会对任务调度长度和处理器利用率造成负面影响，为了改善多核系统对关联任务的处理性能，针对关联任务在多核处理器上的调度特点，提出一种并行感知调度算法。计算各任务与终点间的最长路径值，按照该值的降序来分配任务调度次序，在分配处理器内核时兼顾关联度和任务最早可执行时间，设置最佳匹配评价函数。实验结果表明，与busHEFT和DTSV算法相比，该算法具有更短的任务调度时延、更少的通信量以及更高的处理器利用率。

Parallel Perceptual Scheduling Algorithm for Related Tasks Based on Multi-Core Processors

The communication delay caused by the parallel scheduling of related tasks on multi-core processors has a negative impact on the task scheduling length and processor utilization.In order to improve the performance of multi-core systems in processing related tasks,this paper proposes a parallel perceptual scheduling algorithm based on the scheduling characteristics of related tasks on multi-core processors.The algorithm calculates the value of the longest path from each task to the end point,and assigns the tasks in descending order of the calculated value.The relation and the earliest execution time are both considered when assigning processor cores,and the best matching evaluation function is created.Experimental results show that this algorithm has fewer task scheduling delay,less communication traffic and higher processor utilization compared with busHEFT,DTSV and other algorithms.