Efficient GPU Implementation of the High-Frequency SBR-PO Method

Several key implementation measures are proposed for effectively mapping the shooting and bouncing rays-physical optics (SBR-PO) method to a graphics processing unit (GPU) to accelerate the solution process of scattering problems. Compute unified device architecture (CUDA) and OptiX from NVIDIA are used to reduce computational time for the modeling of monostatic scattering from electrically large and arbitrarily shaped objects. Results obtained using the SBR-PO code on GPU are compared against results from literature and those generated by commercial software. The good agreement and very significant speedup achieved show that the newly developed SBR-PO code on GPU is capable of modeling a very large-size platform for practical rapid design.