基于改进遗传算法的3D NoC测试优化  

TEST SCHEDULING OF 3D NOC UNDER IMPROVED GENETIC ALGORITHM

作　　者：凌景[1] 唐静[1] LING Jing TANG Jing(Chaohu College, Chaohu Anhui 2380)

机构地区：[1]巢湖学院,安徽巢湖238000

出　　处：《巢湖学院学报》2018年第3期72-79,共8页Journal of Chaohu University

基　　金：国家自然科学基金(项目编号:NO.61561012);巢湖学院校级项目(项目编号:XLY-201705;XLZ-201702)

摘　　要：采用将固定的带宽TAM根据需要动态划分为多条子TAM分配给IP核测试数据传输的并行测试策略,重用NoC作为TAM,采用XYZ路由算法,以测试时间作为约束函数,在TAM带宽约束下采用改进的遗传算法为待测IP核选择合适的调度顺序以获得最短测试时间,提高测试效率。实验结果表明,针对不同规模的NoC,使用云模型更新寻优到精英种群后,再使用遗传算法更新个体的方法能找到最优测试方案,减小测试时间,提高资源利用率。This paper adopts the parallel test strategy that the fixed bandwidth TAM is assigned to IP cores dynamically, reuses No C as the test access mechanism and adopts XYZ routing algorithm, test time as constraint function. It chooses appropriate scheduling order for IP core under improved genetic algorithm to get the shortest test time and improve test efficiency under TAM bandwidth constraint. The experiment results show, for different sizes No C, the method that after finding the elite population by using cloud model optimization, uses genetic algorithm to update individual can find the optimal test plan, reduce test time and improve resource utilization.

关 键 词：三维片上网络 带分复用 云模型 遗传算法 

分 类 号：TP306[自动化与计算机技术—计算机系统结构]