Graphics Processing Unit Implementation of Multilevel Plane-Wave Time-Domain Algorithm

被引:0
|
作者
Liu, Yang [1 ]
Yuecel, Abdulkadir C. [1 ]
Lomakin, Vitaliy [2 ]
Michielssen, Eric [1 ]
机构
[1] Univ Michigan, Dept Elect Engn & Comp Sci, Ann Arbor, MI 48109 USA
[2] Univ Calif San Diego, Dept Elect & Comp Engn, San Diego, CA 92093 USA
来源
IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS | 2014年 / 13卷
基金
美国国家科学基金会;
关键词
Graphics processing unit (GPU); multilevel plane-wave time-domain (ML-PWTD) algorithm; NVIDIA CUDA fast Fourier transform library (CUFFT);
D O I
10.1109/LAWP.2014.2350967
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
A graphics processing unit implementation of the multilevel plane-wave time-domain algorithm for rapidly evaluating transient electromagnetic fields generated by large-scale dipole constellations is proposed. The implementation achieves 50x speedup and 60%-75% memory reduction compared to its serial CPU implementation.
引用
收藏
页码:1671 / 1675
页数:5
相关论文
共 50 条