针对FPGA的结构特点,借鉴ASIC布局算法中非线性建模思想,提出一种应用于大规模FPGA的解析式布局算法.该算法以非线性线长为目标,采用较少迭代次数的共轭梯度方法作为求解器,解决组合优化方法时间大量消耗问题.实验结果表明,该方法能够在较短的时间得到较好的布局质量,与FastPlace的结果对比证明了其有效性.
This paper presents an analytical placement algorithm to handle large-scale FPGAs design fast. Motivated by the idea in large ASIC design, the algorithm uses nonlinear wire-length as objective and takes FPGAs special structure in account. A nonlinear programming is solved to get the cell position by a conjugate gradient method with less iterations to reduce the runtime of the combinatorial optimization. Experimental results show that this approach can effectively reduce runtime and improve placement quality. And the final wire-length results compared to FastPlace also testify its reasonable quality.