Fast calculation of computer-generated holography using multi-graphic processing units

A process of generating a digital hologram requires a lot of time-consuming computations. Therefore, it is important to reduce the computation time or the number of computations for achieving real-time digital holographic video generation. In this paper, we propose a method of parallelizing the computations using multiple GPUs with CUDA and OpenMP and an optimization method for reducing the computation time. Through simulation, we verified that the speed of generating digital hologram can be greatly increased by our optimization methods and in proportion to the number of GPUs. As a result, our methods could generate the digital holographic video which has the resolution of 1,920 × 1,024 from about 2,048 light sources at a speed of 31 hologram/s.