GPU acceleration of linear systems for computational electromagnetic simulations

The use of Graphical Processing Units (GPUs) to perform computational electromagnetic simulations has been proven over past several years to increase the calculation speed. By examining the various mathematical processes used in various techniques, appropriate algorithms for the GPU can be developed to speed up the simulations. Understanding how to map algorithms appropriately to the GPU is of vital importance to achieving the maximum gains in speed. This paper will detail a simple technique that can be used to enable the use of GPUs for general linear systems solvers with appreciable speed up factors.