MCM placement problem with GASA multi-objective optimization strategy

Placement of multiple dies on an MCM substrate is a difficult combinatorial task in which multiple criteria need to be considered simultaneously to obtain a true multi-objective optimization. In this paper we described a MCM placement model for the multi-objective optimization problem and solved this model by the simulated annealing SA algorithm and the hybrid optimization strategy GASA (namely the combination of genetic algorithm and simulated annealing) respectively. Our design methodologies consider multi- objective component placement based on thermal reliability, routing length and chip area criteria for multi-chip module. The purpose of the multi-objective optimization placement is to enhance the system performance, reliability and reduce the substrate area by obtaining an optimal cost during multi-chip module placement design phase. For reliability considerations, the design methodology focuses on the placement of the power dissipating chips to achieve uniform thermal distribution. For route-ability consideration, the total wire length minimization is estimated by bounding box approximation method. For substrate area consideration, the area is estimated by minimum area contains all chips. The cost function is formulated by the weight sum calculation. For design flexibility, different weights can be assigned depending on designer´s considerations. Various methods including simulated annealing and hybrid generic algorithm are applied to solve the placement solutions. 3-D finite element analysis (FEA) is carried out to assess thermal distribution in MCM substrate. The optimization results of various weighting assignments obtained by different algorithms are compared. In addition, an auto generated optimal placement layout based on the analytical solution is also presented.