16nm工艺下的新一代静态时序分析技术SOCV  

New-generation Static Timing Analysis Technique SOCV of 16 nm Process

作　　者：胡云生[1] 胡越黎[1,2] 王伟平[1] 承文龙 杨晔晨 

机构地区：[1]上海大学微电子研究与开发中心,上海200072 [2]上海大学机电工程与自动化学院,上海200072

出　　处：《计算机测量与控制》2017年第4期213-215,243,共4页Computer Measurement &Control

摘　　要：工艺偏差在更加先进的工艺节点上别的尤为重要;最初使用工艺偏差方法学(on-chip variation,OCV)使用一个系数因子在整条时序路径上放大缩小来模仿工艺变化,这种方法学过于悲观;先进的片上误差方法学(advanced ocv,AOCV)可以在不同的时序路径上不同的逻辑深度添加不同的系数因子来模拟工艺误差;但是这种方法学分析的时间太长,消耗的内存太多,并且分析的场景出现的概率很低;介绍一种在16nm下最新的一代时序分析技术-统计学片上误差分析(statistic ocv,SOCV);SOCV能够模拟某种误差使得延时出现的概率,因此SOCV较AOCV更为准确,能够去除部分特别悲观和特别乐观的场景;SOCV耗时明显要低于AOCV,因此SOCV能加快sign-off的时间。Process variation on a more advanced process nodes is particularly important. The original methodology on-chip variation （OCV） u sing a coefficient factor zoom on a timing path to mimic the process of change, which is too pessimistic. Advanced on-chip variation can uses the differ ent coefficient factor to simulate the process on different paths and different logical depth. But this analysis consume too long time and too much memo ry, and the analysis scenario has a low probability. 16 nm is presented in the text of the latest generation statistical static timing analysis （SOCV） SOCV can simulates the probability a certain error delay, so SOCV is more accurate than AOCV, and it can remove the pessimistic and optimistic sce narios. SOCV is less than AOCV in timing consumption, so SOCV can speed up the sign off.

关 键 词：静态时序分析 AOCV SOCV 

分 类 号：TN47[电子电信—微电子学与固体电子学]