Studies on DA and Intelligent CAD DB for VLSI Design

各种改进版本的转导方法（最初由Kambayashi和murroga开发）和新的应用，如fpga（现场可编程门阵列）的设计。转导法的优点之一是可以重用设计结果。此属性适用于CAD数据库。提出了一种适用于CAD数据库和用户协调的数据模型。实现了转导方法程序和CAD数据库。结果详情如下：(1)基于误差补偿的转导方法改进：为了提高网络模式修正能力，开发了基于误差补偿的转导方法。我们对该方法进行了改进，改进了计算时间，这是误差补偿的一个严重缺点。(2)采用转导法设计大型电路：由于转导法处理的电路的最大尺寸由BDD（二进制决策图）大小决定，因此我们将选择电路的一部分来改进大型电路的设计。我们已经开发了几种方法来选择这样的部分电路。(3)换能法FPGA设计：对于给定的FPGA电路，通过合并逻辑块并去除冗余连接生成改进电路。(4)数据-知识协调模型的实现：对于CAD数据库来说，重要的是使用分布在不同地点的各种数据库和知识库。我们实现了一个由三个层次组成的系统，一个是数据库，一个是知识库，一个是协调。采用对象代理模型实现。此外，还开发了CAD数据库用户协调的基本机制。

Various improved versions of transduction methods (originally developed by Kambayashi and Muroga) and new applications such as design of FPGAs (Field Programmable Gate Arrays). As one of the advantages of transduction method is that we can reuse design results. This property is suitable for CAD databases. A new data model suitable for CAD database and user coordination is also developed. Programs of transduction methods and CAD databases are implemented. Details of the results are as follows.(1) Improvement of transduction methods based on error-compensation : In order to improve network pattern modification capability, transduction methods based on error-compensation were developed. We made a new version of the method and improved the computation time, which was a serious disadvantage of error-compensation.(2) Design of large circuits using transduction method : As the maximum size of the circuits handled by transduction methods is determined by the BDD (Binary Decision Diagram) size, we will select a part of a circuit to improve the design for a large circuit. We have developed several methods to select such a partial circuits.(3) FPGA design by transduction method : For a given FPGA circuit, an improved circuit is generated by merging logic blocks and removing redundant connections.(4) Realization of a data-knowledge coordination model : For CAD databases it is important to use various kinds of databases and knowledgebases distributed in various locations. We have implemented a system consisting of three hierarchies, one fore databases, one for knowledgebases and one for coordination. Object deputy model is used for realization. Furthermore, fundamental mechanisms for user coordination for CAD databases are developed.