Accelerating the computation of the discrete currents method by modification takes into account the architectural features of a modern PCs

Abstract

We develop a modification of the popular modeling techniques diffraction phenomena in the resonance wavelength range in order to accelerate the calculations at the expense of the opportunities hidden in the architecture of modern PCs. This modification uses the vector registers of the processor. As a result of the modification of the method of discrete currents (DCM), which aims to use the vector registers, combined with the optimization of the work with the cache memory, we got the result, which is well illustrated by the following example. In case of deciding the diffraction problem on 80 screens (wave number 14π) on a 4-core CPU, following the DCM, we got acceleration 30-40 times in comparison with the performance on a single processor core without the use of vector registers and cache processor optimization.