图计算在ATPG中的应用探究  

Research on the application of graph processing in ATPG

作　　者：毛伏兵 彭达 张宇[1,2,3,4] 廖小飞 姜新宇[1,2,3,4] 杨赟[1,2,3,4] 金海 赵进[1,2,3,4] 刘海坤 王柳峥[5] Fubing MAO;Da PENG;Yu ZHANG;Xiaofei LIAO;Xinyu JIANG;Yun YANG;Hai JIN;Jin ZHAO;Haikun LIU;Liuzheng WANG(National Engineering Research Center for Big Data Technology and System,Huazhong University of Science and Technology,Wuhan 430074,China;Services Computing Technology and System Lab,Huazhong University of Science and Technology,Wuhan 430074,China;Cluster and Grid Computing Lab,Huazhong University of Science and Technology,Wuhan 430074,China;School of Computer Science and Technology,Huazhong University of Science and Technology,Wuhan 430074,China;Huawei HiSilicon Co.,Ltd.,Shenzhen 518129,China)

机构地区：[1]华中科技大学大数据技术与系统国家地方联合工程研究中心,武汉430074 [2]华中科技大学服务计算与系统教育部重点实验室,武汉430074 [3]华中科技大学集群与网格计算湖北省重点实验室,武汉430074 [4]华中科技大学计算机科学与技术学院,武汉430074 [5]华为海思半导体有限公司,深圳518129

出　　处：《中国科学：信息科学》2023年第2期211-233,共23页Scientia Sinica(Informationis)

基　　金：国家自然科学基金(批准号:61832006,61825202,62072193);中央高校基本科研业务费资助(HUST)(批准号:2020kfyXJJS018,2020kfyXJJS109)项目。

摘　　要：ATPG(automatic test pattern generation)是VLSI(very large scale integration circuits)电路测试中非常重要的技术,它的好坏直接影响测试成本与开销.然而现有的并行ATPG方法普遍存在负载不均衡、并行策略单一、存储开销大和数据局部性差等问题.由于图计算的高并行度和高扩展性等优点,快速、高效、低存储开销和高可扩展性的图计算系统可能是有效支持ATPG的重要工具,这将对减少测试成本显得尤为重要.本文将对图计算在组合ATPG中的应用进行探究;介绍图计算模型将ATPG算法转化为图算法的方法;分析现有图计算系统应用于ATPG面临的挑战;提出面向ATPG的单机图计算系统,并从基于传统架构的优化、新兴硬件的加速和基于新兴存储器件的优化几个方面,对图计算系统支持ATPG所面临的挑战和未来研究方向进行了讨论.Automatic test pattern generation(ATPG)is a very important technology in very large-scale integration circuit testing,and its quality directly affects the test cost and overhead.However,existing parallel ATPG approaches generally have problems,such as load imbalance,single parallel strategy,expansion of the storage overhead,and poor data locality.Due to the advantages of high parallelism and high scalability of graph computing,graph processing systems that are fast and efficient and have low storage overhead and high scalability may be significant tools to effectively support ATPG,which will be particularly important for reducing test costs.This paper explores the application of graph processing in combined ATPG.We introduce the graph processing model,which transforms an ATPG algorithm into a graph algorithm;analyze the challenges faced by existing graph processing systems applied to ATPG;and propose a stand-alone graph processing system for ATPG.Then,we discuss the challenges and future research directions of graph processing systems supporting ATPG in terms of optimizing traditional architectures,accelerating emerging hardware,and optimizing emerging storage devices.

关 键 词：图计算 超大规模集成电路 自动测试向量生成 电子设计自动化 电路测试 

分 类 号：TN407[电子电信—微电子学与固体电子学] O157.5[理学—数学]