摘要
提出了一种改进的基于粒子群算法的优化布局算法(Improved Particle Swarm Optimization,IPSO)来替换原有的基于模拟退火(Simulated Annealing,SA)算法的优化布局算法,使其更加适用于大型三维片上网络的仿真。通过比较这两种算法的基本思想,给出了这两种算法的实现步骤并详细介绍了IPSO算法的改进思路。最后利用一款现有的三维片上网络仿真器进行了仿真验证。结果表明,提出的IPSO算法比原来的SA算法更适用于大型三维片上网络的仿真。
We presented an improved particle swarm optimization algorithm based algorithm to optimize the floorplan,called improved particle swarm optimization（IPSO）algorithm,and replaced the original floorplan optimization algorithm based on simulated annealing algorithm（SA algorithm）to make it more suitable for large-scale three-dimensional networks-on-chip simulation.We compared the basic idea of these two algorithms.The details of the implementation steps of the two algorithms and improvement ideas of IPSO algorithm were given.We verified the performance improvement using an existing 3D NoC simulator.The simulation results show that the proposed IPSO algorithm is more suitable for large-scale three-dimensional simulation of 3D NoCs than the original SA algorithm.
出处
《计算机科学》
CSCD
北大核心
2015年第7期114-117,124,共5页
Computer Science
基金
国家自然科学基金:基于柏拉图立体多级裂变模型的3D NoC拓扑结构的研究(61272006)资助
