多目标多约束的集成电路统计优化策略

A Multi-Object and Multi-Constraint IC Statistical Optimization Strategy

对于集成电路设计、生产过程中的多目标、多约束统计优化问题,本文提出了“合格率足够高”的优化宗旨,并从概率论的基本原理出发,结合集成电路的特点,导出了一种合格率的近似表述方法,提出的变权重Monte Carlo法编程简便,效率高。采用这些优化策略设计的集成电路合格率优化系统取得了比较好的结果。

For the multiobject and multiconstraint statistical optimization in IC design and production, an optimization object of 'yield high enough' is proposed. From the basic principle of probability,combined with IC characteristics,an approximation expression of yields has been derivied. A changed-wei-ght Monte Carlo method is presented, which is easy to program and highly efficient. Good results have been obtained by the IC yield optimization system designed with the optimization strategy.