Evolutionary approach for solving standard cell placement In VLSI physical design automation

Abstract

The increased complexity in VLSI cell placement due to increase in number of transistors requires the application of various heuristic approaches including evolutionary approaches for obtaining better results. Evolutionary approaches are able to produce high quality placement solutions for standard cell circuits in comparison to other sophisticated algorithms. This work proposes an algorithm named, Evolutionary Approach for Standard Cell Placement problem (EASCP). The proposed algorithm follows a variant of genetic paradigm where a trial value is associated with each candidate solution of the population to improve the quality of solution by replacing the abandoned solutions. With a key feature of exploring the solution space, the proposed EASCP uses a repository for keeping track of the best solutions per generation, thereby, exploiting the solution space. The algorithm was simulated on test circuits. The experimental results obtained a layout with better wirelength in comparison to simple genetic algorithm at the cost of fractional increase in runtime.