Parallel Optimization and Simulation Applications of Electromagnetic Integral Equations on Domestic Heterogeneous Platforms

doi:10.11871/jfdc.issn.2096-742X.2025.05.003

Abstract

Abstract:

[Objective] To fully utilize the computing resources of domestic high-performance platforms and enhance the engineering application capabilities of electromagnetic field numerical simulation algorithms, [Methods] this paper proposes a three-level hybrid parallel architecture of MPI-OpenMP-HIP and a high-efficiency implementation scheme for the method of moments (MoM), a multi-level adaptive cross approximation algorithm (MLACA), and a multi-level fast multipole algorithm (MLFMA) in the electromagnetic field integral equation method by considering the heterogeneous hardware characteristics of the platform. [Results] Through architecture design, task allocation, communication control, and load balancing, the high-efficiency parallel computing of electromagnetic field simulation algorithms have been achieved on a computing cluster configured with distributed nodes, multi-cores, and domestic GPU-like accelerator cards, which can be applied to industrial electromagnetic simulations such as device parameter extraction and target scattering cross-section evaluation. [Conclusions] Experimental results show that the three-level hybrid parallel architecture significantly improves the computing efficiency and application capabilities of electromagnetic simulation algorithms, providing a practical example for the development of new generation electromagnetic simulation software for domestic high-performance computing platforms.

Key words: electromagnetic integral equations, three-level hybrid parallel, domestic heterogeneous platforms

HE Shiquan, WANG Xu, ZHANG Yu. Parallel Optimization and Simulation Applications of Electromagnetic Integral Equations on Domestic Heterogeneous Platforms[J]. Frontiers of Data and Computing, 2025, 7(5): 28-40, https://cstr.cn/32002.14.jfdc.CN10-1649/TP.2025.05.003.

Figures/Tables 15

Fig.1

Table 1

Table 2

Fig.2

Table 3

Fig.3

Fig.4

Table 4

Table 5

Fig.5

Fig.6

Table 6

Fig.7

Table 7

Table 8

References 20

[1]	吕英华. 计算电磁学的数值方法[M]. 清华大学出版社, 2006: 175.
[2]	赖奔. 高阶矩量法及其快速算法的研究与应用[D]. 陕西: 西安电子科技大学, 2011. DOI: 10.7666/d.Y1958755.
[3]	何十全. 非均匀复杂结构目标电磁散射理论建模与高效算法研究[D]. 电子科技大学, 2011. DOI: CNKI:CDMD:1.1011.191823.
[4]	TAMAYO J M, HELDRING A, RIUS J M. Multilevel adaptive cross approximation (MLACA)[J]. IEEE Transactions on Antennas and Propagation, 2011, 59(12): 4600-4608.
[5]	胡俊, 聂在平, 王军, 等. 三维电大目标散射求解的多层快速多极子方法[J]. 电波科学学报, 2004, (5):509-514+524.
[6]	胡俊, 聂在平, 雷霖, 等. 三维局部多层快速多极子算法[J]. 系统工程与电子技术, 2006, (3): 329-330+335.
[7]	伊君翰. 基于多核的并行编程模型[D]. 复旦大学, 2008.
[8]	潘小敏, 盛新庆. 一种高性能并行多层快速多极子算法[J]. 电子学报, 2010, 38(3): 580-584.
[9]	周兰花, 付彬, 李仁发, 等. 基于异构计算的三维FDTD并行算法及其在电磁仿真中的应用[J]. 计算机工程与科学, 2017, 39(7): 1241-1248.
[10]	麻连凤. 积分方程求解复杂目标电磁散射问题的关键技术研究[D]. 电子科技大学, 2011.
[11]	哈林顿. 计算电磁场的矩量法[M]. 国防工业出版社,1981: 8-9.
[12]	王少刚, 关鑫璞, 王党卫, 等. 一种高阶矩量法迭代求解的预处理技术[J]. 微波学报, 2008, (4): 20-23+27.
[13]	陈磊. 基于自适应交叉近似算法快速分析理想导体电磁散射特性[D]. 安徽大学, 2013.
[14]	胡俊. 复杂目标矢量电磁散射的高效方法-快速多极子方法及其应用[D]. 四川: 电子科技大学, 2000.
[15]	HE P, LIU J, GU H, et al. Modified Born series with virtual absorbing boundary enabling large-scale electromagnetic simulation[J]. Communications Physics, 7(1): 1-8.
[16]	XIE W, WEI Y, PAN D, et al. Antenna electromagnetic parameters prediction based on data augmentation and neural network[C]. 2024 IEEE International Workshop on Radio Frequency and Antenna Technologies (iWRF&AT), Shenzhen, China, 2024: 462-466.
[17]	LI C Y, ZHANG T C, BAO H G, et al. Accelerated multi-physics analysis of electromagnetic energy selective surfaces with a space mapping algorithm[J]. Engineering analysis with boundary elements, 2023,155: 140-147.
[18]	ZHU M T, LU H Z, TONG M S. An Efficient GPU Acceleration Scheme for Solving Electromagnetic Problems with Moderate Scales[C]. 2024 International Conference on Electromagnetics in Advanced Applications (ICEAA), Lisbon, Portugal, 2024: 548-550.
[19]	王俊午. 电磁积分方程快速直接求解法的研究与应用[D]. 电子科技大学, 2021.
[20]	孙洁. 用于超电大散射计算MLFMA并行化及其关键问题的研究[D]. 电子科技大学, 2014.

类别	LU分解（s）	总时间（s）	电阻R（mΩ）	电感L（nH）
CPU单核	768	776	56.98	96.67
CPU32核	28	36	57.03	96.67
单加速卡	12	20	57.05	96.67
Maxwell	—	12	59.75	101.54

类别	LU分解（s）	总时间（s）
CPU32核比单核加速比	27.4	21.5
单卡比CPU单核加速比	64	38.8
单卡比CPU32核心加速比	2.3	1.8

	总计算时间（s）	电阻（mΩ）	电感（nH）
MLACA	88	0.759	4901.8
Maxwell	55	0.855	5170.0

计算资源配置（节点×HIP核数）	并行核数	每迭代步计算时间（s）	并行效率	加速比
25×4,096	102,400	290.34	基准	基准
50×4,096	204,800	158.17	91.77%	1.84
125×4,096	512,000	95.16	61.01%	3.05
250×4,096	1,024,000	57.94	50.10%	5.01

计算资源配置（节点×HIP核数）	并行核数	每迭代步计算时间（s）	并行效率	加速比
50×1,024	51,200	167.34	基准	基准
50×2,048	102,400	159.39	52.49%	1.05
50×4,096	204,800	158.17	26.45%	1.06