基于多核系统NOC架构的静态列表调度算法*

在单芯片多核系统中,NoC已成为主流片上通信架构,有效的任务调度是挖掘计算并行性的重要方面。本文在经典静态列表调度基础上,针对HEFT算法中节点排序会得出较多的优先级相同节点的问题,提出一种节点二次排序的调度方法,在边的调度上应用了ALAP原则,改进算法有效提高了调度效果。实验表明：新方法对bl、blcomp、blio等节点优先级算法得出的任务列表均有良好的调度效果,适应性较好；对于2D MESH同构NoC平台,改进算法对三种节点优先级算法有1.15倍的平均加速比,最大可有1.27倍加速比。

A static list scheduling algorithm for NoC

NoC becomes the main communication architecture of MPSOC system. As its hardware being mature, the scheduling technology for NoC becomes the key factors to restrict the performance of the multi-core. It has been becoming the hot point to research the more appropriate scheduling algorithm to make full parallelism of multi-core. Considering the efficiency of classical HEFT algorithm decreased by many nodes with the same priority, this paper presents a novel scheduling algorithm to gain the better scheduling result by ranking those same priority nodes again and application of the rule of ALPA in edges scheduling. The proposed algorithm has good applicability to schedule lists derived from a variety of node priority. The experiment result shows that the average and the maximum speedup compared with the HEFT algorithm is 1.15 and 1.27.