“多塔”3D-SIC的量子粒子群测试调度方法  

A Test Scheduling Method for Multi-Tower 3D-SIC Based on Quantum-Behaved Particle Swarm Optimization Algorithm

作　　者：崔小乐[1] 王文明[1] 缪旻[2] 金玉丰[1] Cui Xiaole;Wang Wenming;Miao Min;Jin Yufeng(The Key Laboratory of Integrated Microsystems, Shenzhen Graduate School of Peking University, Shenzhen 518055;School of Information Communication Engineering, Beijing Information Science & Technology University, Beijing 100192)

机构地区：[1]北京大学深圳研究生院集成微系统科学工程与应用重点实验室,深圳518055 [2]北京信息科技大学信息与通信工程学院,北京100192

出　　处：《计算机辅助设计与图形学学报》2017年第1期196-210,共15页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家"九七三"重点基础研究发展计划项目(2015CB057201);广东省自然科学基金(2015A030313147);广东省科技计划项目(2014B090913001);长城学者培养计划(CIT&TCD20150320)

摘　　要：针对"多塔"结构的3D堆叠集成电路(3D-SIC)测试耗时很长的问题,提出一种基于量子粒子群优化的测试调度方法,以缩短测试时间.首先,构造初始粒子群用以表示初始可行解,产生具有量子行为的新粒子,并更新粒子群;然后进行粒子群的迭代进化以获取全局最优解.最小化"终堆"测试时间和集成过程总测试的调度结果均表明,该方法可显著地缩短测试时间;当复杂晶片集成在3D-SIC底层时,"终堆"测试时间较短,而集成过程的总测试时间较长.Multi-tower three dimensional stacked integrated circuits(3D-SIC)have a prolonged test time.This work proposes a test scheduling method of3D-SIC based on quantum-behaved particle swarm optimization(QPSO)to reduce the test time.Firstly,it constructs an initial particle swarm that represents the feasiblesolutions;then it generates the new quantum-behaved particle and updates the particle swarm based onthe fitness values and the constraints;finally,it approximates the optimal scheduling solution till the end ofthe swarm evolution.The experimental results on the final stack and the integration process show that thismethod reduces the test time of3D-SIC obviously.The test time is shorter for the stacked final3D-SIC,andvice versa for the3D-SIC integration process,if the complex die is integrated as the bottom die.

关 键 词：“多塔” 3D-SIC 量子粒子群优化 测试调度 “部分堆” “终堆” 

分 类 号：TN47[电子电信—微电子学与固体电子学]