Analytic annealing for macrocell placement optimization

An adaptive procedure is presented for the macrocell placement problem. The procedure utilizes bivariate formulation for optimizing the placement of macros (modules) considering their geometrical specifications. By minimizing the dependence of the optimal solutions on the initial placement and the order of movement of modules, better designs are possible with fewer trials. Computer implementation of the procedure is described and its performance is compared with two adaptive heuristics; a simulated annealing heuristic and a heuristic based on a genetic algorithm. For the given test problem, the presented procedure performed better as compared to the two heuristics. Also, for two other test problems, the results are presented and compared with the results obtained by two published programs, one based on a heuristic procedure and the other based on an analytical technique. The results clearly demonstrate the improved performance of the analytic annealing procedure for macrocell placement optimization.