电子学报 ›› 2017, Vol. 45 ›› Issue (3): 740-746.DOI: 10.3969/j.issn.0372-2112.2017.03.034

• 学术论文 • 上一篇    下一篇

基于聚类的异步时钟SoC测试

凌立1, 江建慧1, 张颖1, 王真2   

  1. 1. 同济大学软件学院, 上海 201804;
    2. 上海电力学院计算机科学与技术学院, 上海 200090
  • 收稿日期:2015-10-28 修回日期:2016-05-19 出版日期:2017-03-25
    • 通讯作者:
    • 江建慧
    • 作者简介:
    • 凌立 男,1988年出生.现为同济大学博士研究生.主要研究方向为VLSI测试与容错计算.E-mail:072506@tongji.edu.cn;张颖 男,1984年出生,现为同济大学副教授.主要研究方向为嵌入式系统设计、处理器的程序自测试.E-mail:yingzhang@tongji.edu.cn;王真 女,1984年出生,博士、高级工程师.现为上海电力学院讲师.主要研究方向为容错计算及计算机系统性能评估.E-mail:wangzhenqq@hotmail.com
    • 基金资助:
    • 国家自然科学基金 (No.61432017)

Clustering-Based Asynchronous Clock Periods SoC Testing

LING Li1, JIANG Jian-hui1, ZHANG Ying1, WANG Zhen2   

  1. 1. School of Software Engineering, Tongji University, Shanghai 201804, China;
    2. College of Computer Science and Technology, Shanghai University of Electric Power, Shanghai 200030, China
  • Received:2015-10-28 Revised:2016-05-19 Online:2017-03-25 Published:2017-03-25
    • Supported by:
    • National Natural Science Foundation of China (No.61432017)

摘要:

为进一步减少片上系统(System-on-Chip,SoC)测试耗时、降低测试成本,本文结合异步时钟测试机制,提出一种基于聚类的测试调度方法.该方法利用了SoC各测试的特征以及异步时钟测试的特点,对测试数据进行预处理.在ITC'02基准SoC集上,将本文方法与未采用异步时钟机制以及基于混合整型线性规划模型求解的方法进行对比.结果表明,本文的方法分别能平均减少测试耗时20.39%和5.53%,提升了调度算法的优化效率.并且在功耗约束较强时,最终调度结果与耗时下界仅相差0.9%.

关键词: SoC测试调度, 异步时钟, 混合整型线性规划模型, 聚类

Abstract:

The clustering-based SoC test scheduling algorithm combined with the asynchronous clock periods testing is proposed to further reduce the SoC(System-on-Chip) test application time (TAT) and test cost.The scheduling algorithm pre-processes the test data by exploiting the characteristics of the tests.After conducting experiments on the ITC'02 SoC benchmark,we find out that the proposed scheduling method based on clustering can reduce TAT by 20.39% and 5.53% on average,when comparing with a synchronous clock testing method and an asynchronous method based on the MILP model,respectively.Besides,when the power constraint is tight,there is only a difference of 0.9% between the scheduling result and the lower bound.

Key words: SoC test scheduling, asynchronous clock, MILP model, clustering

中图分类号: