基于PSO的粗颗粒度可重构处理器时域划分算法设计

A PSO Temporal Partitioning Algorithm used in the Compiler of Reconfigurable Processor

微粒群算法(PSO)是一种新颖的群智能仿生进化优化算法,它简单、可控性强、易实现且具有很强的优化能力。本文首次将PSO算法引入时域划分领域,给出了一种基于PSO的时域划分优化算法PSOTP,用于粗粒度可重构系统任务编译过程中的时域划分优化。PSOTP算法运用数据流图(DFG)节点的序列作为微粒的位置,微粒中节点位置改变后两个序列的差分作为速度,以任务划分后的子模块数、数据通信量和可重构计算资源的面积利用率作为优化目标,是一种基于权重的多目标优化算法。实验表明,PSOTP算法在划分结果的性能上明显优于传统的ASAP、ALAP和表调度算法以及基于权重的表调度(Priority-List,PL)算法,同时与基于模拟退火遗传算法SAGA的时域划分算相比,可以用更少的迭代次数取得相当的优化效果,运算速度也更快。

Particle Swarm Optimization （PSO） is a new optimization algorithm based on Group Intelligence of bionic evolution. It is easy and controllable, and has a strong ability in continuous space optimization. A novel temporal partitioning algorithm based on PSO is proposed in this paper, to solve the optimization problem in temporal partitioning of Reconfigurable System REmus. It consid- ers three main factors in partitioning： Sub-module Numbers, Communication Cost and Area Efficiency. Simulation Results show that this new method significantly increased the performance of temporal partitioning on REmus system.