Preprints IPM . M. In. Keldysh . 2013. No. 16 . 20 s. The algorithm for solving grid equations of an explicit difference scheme for Maxwell's equations is formulated in a vectorized form. The computing module is based on Cuda and OpenMP technologies with optimization of access to graphics card memory. Vector operations are performed on graphics processors. Operations that require random access to memory are performed by the central processor in multithreaded mode. The use of graphics cards in the nodes of the cluster provided an acceleration of calculations compared to the traditional parallel architecture. The calculation is accelerated by 11 times for tasks with a small amount of RAM and by 2-3 times for other tasks.