FPGA连线连接盒中基于信息熵优化的结构设计

Optimal Design of Topological Structure for FPGA Connection Box Based on Information Entropy

连线连接盒（CB）的结构设计是FPGA结构设计中的重要部分，从工程应用的角度提出一种适用于任意可编程开关比例的CB设计方法．首先提出一种新颖的按列移行方法，用于给出特定约束下可编程开关的初始布局；再以可编程开关在CLB输入引脚上分布的最大信息熵为优化目标，利用模拟退火算法得到优化的CB结构．实验结果表明，与其他方法相比，该方法提高了约8％的布通率，并降低了2％～6％的面积延时积．

Optimum design of topological structure for connection box （CB） is an important issue in FPGA. We propose a design method that is applicable to any proportion of switches. Firstly, a novel shifting row by column method is used to initialize the placement of the switches under certain constrains. Then the simulated annealing method is employed to maximize the information entropy of the switch distribution for the CLB input pins. The experimental results show that the proposed method obtains nearly an 8% increase in routability of CB. Moreover, it achieves a 2%-6% reduction in area-delay product.