Open Access Research on an Improved Algorithm for 3D NoC Floorplanning Based on Particle Swarm Optimization

Abstract: Nowadays, three-dimensional network-on-chip (3D NoC) with its shorter global interconnects, higher perform-ance, lower loss of interconnection, higher packing density, smaller volume, and many other advantages, has drawn more and more attention in both industrial and academic circle. In this paper, an improved algorithm, named the algorithm based on particle swarm optimization algorithm to optimize the floorplans (PSO-NoC), has been proposed with simula-tions conducted to rectify this algorithm. The simulation results are compared with the original Simulated Annealing-NoC. The experimental results show that the PSO-NoC algorithm reduces the latency and improves the throughput com-pared with the original one. Particularly, the CPU’s process time is significantly decreased by 52.78 % in the average case and 87.08 % in the best case respectively