Research Description:

The reuse of well-tested and optimized design objects allows decreasing design times, increasing design quality, and improving the predictability of designs. Reuse reaches from selecting cells from a library up to adapting already designed objects. In this project, we investigate the reuse of designed objects in CAD frameworks. To a given requirement specification, we try to find suitable design objects in a design database. Suitable means that the found objects are similar to the given requirement specification and, therefore, candidates to solve the current design problem by instantiation or by (small) adaptations.

Based on experiences in other disciplines, mainly in software engineering and case-based reasoning, we developed a feature-based model to describe design objects and requirement specifications. Here, design objects and specifications are characterized by their properties (features). Based on the similarities of features we compute the similarities of design objects and specifications which roughly expresses the expense of adaptation. Our model considers multi-functional units as well as generic, e.g. parameterized, modules. To handle dependences of different features, we propose the usage of rules.

With our prototype implementation RODEO, we examined our model. RODEO searches for a given requirement specification the most suitable modules in a design database. The designer has the possibility to weight the properties of the specification. Our implementation is suitable to perform interval search in a cell library as well as to retrieve design objects for adaptation. After the search, an explanation component informs the designer about necessary adaptation steps. RODEO is integrated in our VLSI design system PLAYOUT.


Staff Members

Joachim Altmeyer

Bernd Noll
RalfSchäfer

Former Members:

Matthias Halbritter
Stefan Ohnsorge

Further Information


Temporary maintenance by Joachim Altmeyer