论文部分内容阅读
本文提出一个基于Kohonen自组织神经网络的以关键路径时延最小为优化目标的时延驱动布局算法。算法的关键是建立面向线网的样本矢量。与面向单元的样本矢量相比,面向线网的样本矢量不仅可以直接处理多端线网,而且能够描述时延信息。实验结果表明,这是一种有效的方法。
This paper presents a delay driven layout algorithm based on Kohonen self-organizing neural network with the objective of minimizing the critical path delay. The key to the algorithm is to create a vector of samples oriented to the net. In contrast to cell-oriented sample vectors, mesh-oriented sample vectors can not only process multi-end nets directly, but also describe delay information. Experimental results show that this is an effective method.