Hardware acceleration of an FMM-FFT solver using consumer-grade GPUs

A surface method of moment (MoM)-based computational electromagnetics solver accelerated via the fast multipole method combined with the fast Fourier transform (FMM-FFT) is studied using different grade GPUs and multi-core CPUs as the hardware acceleration platforms. Economics related to hardware choice is studied and consumer grade GPUs are found to be very competitive against the high-end GPUs when employed in conjunction with multi-core CPUs. Though the code is equally capable of predicting radiation and scattering, this work concerns with predicting Radar Cross Section (RCS) of electrically large objects.