Kohonen神经网络在时延驱动布局中的应用
APPLICATION OF SELF-ORGANIZING NEURAL NETWORK IN TIMING-DRIVEN PLACEMENT
-
摘要: 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。Abstract: In this paper, a timing-driven placement algorithm using Kohonen self-organizing is proposed. The object function of the algorithm is to minimize the critical nets. In the algorithm, net-based similarity vector is presented. Compared with the cell-based similarity vector, net-based similarity vector not only can directly deal with the multi-terminal nets, but also can describe the information about timing. The experimental result shows that it is a effective method.
-
Sriram M, Kang S M. A modified Hopfield network for the two-dimensional module placement. Proc. of ISCAS, New Orleans Louisiana: 1990, 1664-1667.[2]Unaltuna M K, Pitchumani V. Quadrisectioning based placement with a normalized mean field network. Proc. of ISCAS, USA: 1993, 2047-2050.[3]Hemani A, Postula A. Cell placement by self-organization[J].Neural Networks.1990, 3(1):377-383[4]沈涛,等.一种可用于布局的人工神经网络.电子学报,1992, 20(10): 100-105.[5]Gloria A D, Faraboschi P, Olivieri M. Block placement with a Boltzmann machine. IEEE Trans. on CAD，1994,13(6): 694-701.[6]焦李成.神经网络系统理论,西安:西安电子科技大学出版社,1995，第2章.[7]Zhang C X. Timing heat and area driven placement using self-organizing semantic maps. Proc. of ISCAS. USA: 1993. 2067-2070. -
计量
- 文章访问数: 1906
- HTML全文浏览量: 95
- PDF下载量: 440
- 被引次数: 0
下载:
