作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2010, Vol. 36 ›› Issue (24): 278-280. doi: 10.3969/j.issn.1000-3428.2010.24.100

• 开发研究与设计技术 • 上一篇    下一篇

基于GPU的FDTD麦克斯韦方程快速求解

邵 桢a,蔡红星b,徐春凤a   

  1. (长春理工大学 a. 计算机科学与技术学院;b. 理学院,长春 130022)
  • 出版日期:2010-12-20 发布日期:2010-12-14
  • 作者简介:邵 桢(1974-),女,讲师、硕士,主研方向:计算机仿真,智能算法;蔡红星,副教授、博士;徐春凤,讲师

Solving Quickly of Finite Difference Time Domain Maxwell’s Equations Based on GPU

SHAO Zhen a, CAI Hong-xing b, XU Chun-feng a   

  1. (a. School of Computer Science and Technology; b. School of Science, Changchun University of Science and Technology,Changchun 130022, China)
  • Online:2010-12-20 Published:2010-12-14

摘要:

采用图形处理器(GPU)为主计算核心,应用时域有限差分法(FDTD)实现电磁学中麦克斯韦方程组的快速求解。通过对FDTD求解麦克斯韦旋度方程的直接时间域的分析,给出FDTD的仿真算法。根据GPU能高效地提高FDTD的仿真速度,解决FDTD仿真算法中的计算量庞大问题。利用GPU在FDTD计算中的处理能力,实现了更长的脉冲持续时间和庞大的模型求解与仿真,在适当的时间内完成了超大量的仿真计算。根据在CPU和FDTD上的实际计算结果表明,基于GPU的FDTD仿真算法具有高精度和高效率等特点。

关键词: 时域有限差分法, 图形处理器, 麦克斯韦方程, 电磁场仿真

Abstract:

Using Graphics Processing Units(GPU) as main computational core, the Finite Difference Time Domain(FDTD) is presented for solving electromagnetic Maxwell’s equations. The FDTD simulation algorithm is developed by analyzing the direct time-domain solution like FDTD Maxwell’s curl equations. It is analyzed and compared with GPU that how GPU can be used to greatly speed up FDTD simulations. So enormous computation problem in FDTD simulations is resolved. Using GPU processing power in FDTD calculations, it implements much longer pulse lengths and larger models, and completes expensive simulations computation in reasonable time. According to the comparing of actual computation results in CPUs and FDTD, It is proved that FDTD simulations based on GPU is accurate and efficient.

Key words: Finite Difference Time Domain(FDTD), Graphics Processing Units(GPU), Maxwell’s equations, electromagnetic field simulation

中图分类号: