Benefits and Challenges of GPU Accelerated Electromagnetic Solvers from a Commercial Point of View
Keywords:
Commercial Solvers, CUDA, FDTD, FEKO, FEM, GPGPU, GPU Acceleration, MoM, RL-GO, SBRAbstract
This paper discusses the benefits but also challenges of GPU accelerated electromagnetic solvers from a commercial point of view, namely using FEKO as example. Specifically, the effects of some of the complex interdependencies between different components are presented. It is shown that despite the advances made in the field of GPGPU computing, and impressive speedups for parts of a program or simplified problems, there are a number of factors to consider before these techniques can be applied to a commercial product that is expected to be robust and, most importantly, to always give trustworthy results for a wide variety of problems.
Downloads
References
D. B. Davidson, Computational Electromagnetics for RF and Microwave Engineers, 2nd ed., Cambridge: Cambridge University Press, 2011.
D. B. Kirk and W. W. Hwu, Programming Massively Parallel Processors – A Hands-on Approach, Burlington: Morgan Kaufmann, 2010.
E. Lezar and D. B. Davidson, “GPU-accelerated methods of moments by example: monostatic scattering,” IEEE Antennas and Propagation Magazine, vol. 52, no. 6, pp. 120-135, Dec. 2010.
M. J. Inman, A. Z. Elsherbeni, and C. J. Reddy, “CUDA based GPU solvers for method of moment simulations,” Annual Review of Progress in Applied Computational Electromagnetics, Tampere, Finland, Apr. 2010.
E. Lezar and D. B. Davidson, “GPU-based Arnoldi factorization for accelerating finite element eigenanalysis,” International Conference on Electromagnetic in Advanced Applications (ICEAA), Torino, Italy, Sept. 2009.
A. Dziekonski, A. Lamecki, and M. Mrozowski, “On fast iterative solvers with GPU acceleration for finite elements in electromagnetics,” 10th International Workshop on Finite Elements for Microwave Engineering, Mill Falls, NH, USA, Oct. 2010.
K. E. Spagnoli, An Electromagnetic Scattering Solver Utilizing Shooting and Bouncing Rays Implemented on Modern Graphics Cards, ProQuest, 2008.
Y. Tao, H. Lin, and H. Bao, “GPU-based shooting and bouncing ray method for fast RCS prediction,” IEEE Trans. Antennas Propagat., vol. 58, no. 2, pp. 494-502, 2010.
ICL, University Tennessee, Knoxville, “MAGMA: Matrix Algebra on GPU and Multicore Architectures,” 2015. [Online]. Available: http://icl.cs. utk.edu/magma/index.html
E. Lezar and U. Jakobus, “GPU accelerated electromagnetic simulations with FEKO,” International Supercomputing Conference, Hamburg, June 2012.
E. Lezar, U. Jakobus, and S. Kodiyalam, “GPU related advances in the FEKO electromagnetic solution kernel,” International Conference on Electromagnetics in Advanced Applications (ICEAA), Torino, Italy, Sept. 2013.
Altair Development S.A. (Pty) Ltd, “FEKO – Field Computations Involving Bodies of Arbitrary Shape,” 2016. [Online]. Available: www.altairhyperworks. com/feko
