【24h】

A genetic approach for Graph Partitioning. An Application to Multi-FPGA systems

机译:图分区的遗传方法。在多FPGA系统中的应用

获取原文
获取原文并翻译 | 示例

摘要

Graph partitioning techniques have been used in a wide range of applications. In some VLSI problems, circuits are represented as graphs and upon achieving the partitioning step, it is necessary to preserve the main structure of the circuit. Most the previous research does not preserve the structure of the graph or uses a difficult encoding. All the previous genetic approaches based their representation on vertices. In this paper we have used the spanning tree edges to represent the solutions to graph partitioning problems. With our solution we preserve the main structure of the circuit and is possible to solve the multi-way partitioning problem with an direct encoding. We have checked our method with the Multi-FPGA partitioning problem. In addition we use the fitness function evaluation step to solve the placement process. The experimental results show how our method can successfully solve the partitioning and placement problems.
机译:图分区技术已被广泛应用。在某些VLSI问题中,电路以图形表示,并且在完成分区步骤后,有必要保留电路的主要结构。以前的大多数研究都没有保留图形的结构或使用了困难的编码。以前所有的遗传方法都基于顶点表示。在本文中,我们使用了生成树边缘来表示图分区问题的解决方案。利用我们的解决方案,我们可以保留电路的主要结构,并且可以通过直接编码来解决多路分配问题。我们已经用Multi-FPGA分区问题检查了我们的方法。此外,我们使用适应度函数评估步骤来解决放置过程。实验结果表明我们的方法如何成功解决分区和放置问题。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号