Design space exploration using the genetic algorithm

A typical VLSI layout problem involves the simultaneous optimization of a number of competing criteria. Rather than generating a single compromise solution, some recent approaches explicitly explores the design space and outputs a set of alternative solutions, thereby providing explicit information on the possible tradeoffs. This paper discuss the use of genetic algorithms (GAs) for design space exploration and propose a solution set quality measure needed to evaluate the performance of set-generating algorithms. A GA for building-block placement is evaluated using the proposed measure and promising results are obtained.