基于混合优化映射算法的NoC自动生成方法研究  

Research on Network-on-Chip Synthesis Flow Based on Hybrid Optimization Mapping Algorithm

作　　者：李君艺[1] 刘怡俊[2] 乐建亮[2] 

机构地区：[1]广东理工职业学院工程技术系,广东广州510091 [2]广东工业大学计算机学院,广东广州510006

出　　处：《系统仿真学报》2017年第5期1141-1146,共6页Journal of System Simulation

基　　金：广东省科技项目(2015B090901060;2016B 090918126;2016B090904001;2016B090903001);2015东职院科研基金重点项目(2015a08)

摘　　要：针对片上网络映射算法中,粒子群优化算法对于离散的优化问题处理不佳,容易陷入局部最优问题,提出了一种基于粒子群优化(particle swarm optimization,PSO)算法和遗传算法(genetic algorithm,GA)的混合优化映射算法(PSO_GA)。选择两个种群分别进行GA和PSO操作,由GA算法中的优良个体代替PSO算法中的初始随机粒子,保留优良粒子的同时,又维持了群体的多样性并提高搜索效率。基于NS-2仿真实验结果表明,采用混合优化映射算法的自动生成工具得出的片上网络对比同等计算规模下的随机映射方式,在网络延迟、吞吐量、链路带宽等方面有明显的优化。To solve the problem of particle swarm optimization algorithm, which is easy to fall into local optimum in network on chip mapping, a hybrid optimization mapping Algorithm based on particle swarm optimization and genetic algorithm was proposed. It implemented GA and PSO separately with two groups, by the superior individuals from GA algorithm instead of the initial random particles from PSO algorithm, which not only maintained the diversity of the group but also improved search efficiency. Simulation results based on NS-2 show that the Network-on-Chip from the automatic generation tools have a good performance in network latency, throughput, and link bandwidth optimization comparing the random mapping under the same amount of computation scale.

关 键 词：2D MESH 混合算 NOC映射 NS-2 

分 类 号：TP273.5[自动化与计算机技术—检测技术与自动化装置]