基于云自适应遗传算法的NoC映射研究  

Research of NoC mapping based on adaptive genetic algorithm of cloud

作　　者：许川佩[1] 陈征南[1] 任智新[1] 

机构地区：[1]桂林电子科技大学电子工程与自动化学院,广西桂林541004

出　　处：《计算机工程与应用》2012年第36期70-74,104,共6页Computer Engineering and Applications

基　　金：国家部委预研基金资助项目

摘　　要：NoC映射是NoC设计中的重要步骤,映射结果的优劣对NoC的QoS约束和通信功耗有着很大的影响。提出一种采用云自适应遗传算法实现NoC映射的方案,该算法利用云模型对传统遗传算法加以改进,以此新方法自动调整遗传算法过程中的交叉概率和变异概率,从而达到优化遗传算法的目的。结合NoC映射中的具体问题,在功耗和延时约束的限制条件下,建立了延时约束下的NoC映射功耗数学模型。实验表明,该方法在NoC映射中取得了良好的效果,降低了通信功耗。NoC mapping is an important step of the NoC design. The quality of the NoC mapping determines the effect of QoS constraints and communication power consumption. This paper proposes a NoC mapping solutions based on cloud adaptive genetic algorithm. The algorithm is based on cloud model to improve the traditional genetic algorithm, taking a new method to improve crossover probability and mutation probability. Combined with the practical problems of NoC mapping, in constraints of power consumption and delay conditions, the power consumption of the mathematical model under the delay restriction of the NoC mapping is established. The experiment results show that the method has obtained favorable effect and reduced the power consumption of communication.

关 键 词：片上网络(NoC) 映射 云模型 遗传算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]