Using MATLAB’s Parallel Processing Toolbox for Multi-CPU and Multi-GPU Accelerated FDTD Simulations
Keywords:
FDTD, MATLAB, multi-cores, multiGPUs, parallel computingAbstract
MATLAB is a good testbed for prototyping new FDTD techniques as it provides quick programming, debugging and visualization capabilities compared to lower level languages such as C or FORTRAN. However, the major disadvantage of using MATLAB is its slow execution. For faster simulations, one should use other programming languages like Fortran or C with CUDA when utilizing graphics processing units. Development of simulation codes using these other programming languages is not as easy as when using MATLAB. Thus the main objective of this paper is to investigate ways to increase the throughput of a fully functional finite difference time domain method coded in MATLAB to be able to simulate practical problems with visualization capabilities in reasonable time. We present simple ways to improve the efficiency of MATLAB simulations using the parallel toolbox along with the multi-core central processing units (CPUs) or the multiple graphics processing units (GPUs). Native and simple MATLAB constructs with no external dependencies or libraries and no expensive or complicated hardware acceleration units are used in the present development.
Downloads
References
A. Z. Elsherbeni and V. Demir, The FiniteDifference Time-Domain Method for Electromagnetics with MATLAB® Simulations. 2nd edition, Edison, NJ: Scitech Publishing, 2015.
M. J. Inman, A. Z. Elsherbeni, J. G. Maloney, and B. N. Baker, “GPU based FDTD solver with CPML boundaries,” in 2007 IEEE Antennas and Propagation Society International Symposium, pp. 5255-5258, 2007.
K. Hayakawa and R. Yamano, “Multi-core FPGA execution for electromagnetic simulation by FDTD,” in 2015 2nd International Conference on Information Science and Control Engineering, pp. 829-833, 2015.
X.-M. Guo, Q.-X. Guo, W. Zhao, and W. Yu, “Parallel FDTD simulation using NUMA acceleration technique,” Progress In Electromagnetics Research Letters, vol. 28, pp. 1-8, 2012.
T. Nagaoka and S. Watanabe, “Multi-GPU accelerated three-dimensional FDTD method for electromagnetic simulation,” in 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, pp. 401-404, 2011.
M. R. Zunoubi, J. Payne, and M. Knight, “FDTD multi-GPU implementation of Maxwell’s equations in dispersive media,” presented at the Optical Interactions with Tissue and Cells XXII, vol. 7897, p. 78971S, 2011.
J. E. Diener and A. Z. Elsherbeni, “FDTD acceleration using MATLAB parallel computing toolbox and GPU,” vol. 32, p. 6, 2017.
W. Shao and W. McCollough, “Multiple-GPUbased frequency-dependent finite-difference time domain formulation using MATLAB parallel computing toolbox,” Progress In Electromagnetics Research M, vol. 60, pp. 93-100, 2017.
“MATLAB multicore,” [Online]. Available: https:// www.mathworks.com/discovery/matlab-multicore. html. [Accessed: 10-Jan-2019].
“Parallel MATLAB: Multiple processors and multiple cores,” [Online]. Available: https://www. mathworks.com/company/newsletters/articles/paral lel-matlab-multiple-processors-and-multiple-cores. html. [Accessed: 10-Jan-2019].
P. F. Baumeister, T. Hater, J. Kraus, D. Pleiter, and P. Wahl, “A performance model for GPUaccelerated FDTD applications,” in 2015 IEEE 22nd International Conference on High Performance Computing (HiPC), pp. 185-193, 2015.
V. Demir and A. Elsherbeni, Computational Electromagnetics Simulator (CEMS) Software Package. 2010.
