Genetic Algorithms for Physical VLSI Design

G.N. Reddy and A. Bhattacharjee (USA)


Physical VLSI design; Genetic algorithms; Modeling,Simulation, and Optimization.


In this paper we presents modeling, simulation, and optimization of physical VLSI design problems using a package of genetic algorithms. This software package demonstrates efficient way to solve physical VLSI design problems using genetic algorithms. Many of the physical design problems are combinatorial optimization problems that are NP-complete. The physical design of VLSI chips involve a sequence of very complex combinatorial optimization problems that are insolvable through exhaustive search algorithms. This include partitioning, floor-planning, placement, global routing, detailed routing, and peak-power estimation. The newly emerging Genetic Algorithms (GAs) are most attractive for such applications. A genetic algorithmic approach for solving the VLSI physical design problems is presented in this paper. Solving of the three basic optimization tasks in physical VLSI design namely; partitioning, automatic placement, and global routing; are presented here. We compared the results with existing alternative algorithm to support the validation our results. The software tool we developed is an user-friendly application that has Graphic User Interface (GUI). We developed it using Borland C++ Builder version 5.0. Interested readers may download the software from the author's web-site:

Important Links:

Go Back