光电MCM分割算法的改进  

Improving on Algorithms for Opto-Electronic MCM Partitioning

作　　者：周正伟[1] 陈伟元[1] 王豪才[1] 

机构地区：[1]电子科技大学CAE中心,四川成都610054

出　　处：《电子学报》2001年第12期1632-1634,共3页Acta Electronica Sinica

基　　金：军事电子预研基金资助 (No .JDA5962 )

摘　　要：芯片级分割技术可用于解决光电MCM(OEMCM)物理设计所面临的散热、系统速度及最大光互连距离等问题 .本文给出了光电MCM分割的模型 ,将经改进的遗传算法 (Gas)用于光电MCM的分割问题中 .改进算法较标准遗传算法更适合光电MCM分割 ,其应用可使系统功耗降低约 5 0 % .Chip-level partitioning technology is used for solving the problems coming from the physical design of Opto-electronic multichip module (OE-MCMs): heat dissipation, system speed, and maximum optic interconnect distance. The physical models and algorithms for OE MCM partitioning are given. This paper uses genetic algorithms (Gas) for opto-electronic system partitioning and improving the genetic algorithm. The modified algorithm is more suitable for OE MCM partitioning than standard Gas, and it can decrease the power consumption about 50%.

关 键 词：光电MCM分割算法 多芯片组件 遗传算法 

分 类 号：TN4[电子电信—微电子学与固体电子学]