Application of the alternating triangular iterative method to solving shallow water hydrodynamics problems with graphics accelerators

Forecasting environmental disasters, both artificial and natural, is currently based on advances in simulation. Fast forecasting is very difficult without the use of parallel computing and supercomputer technologies. The large volume of information to be processed and the complexity of calculations require to use of computing clusters with graphics cards to increase the computing performance and the data processing rate. This paper deals with the development of a software module in CUDA C to simulate the hydrodynamic processes in shallow water bodies, including the solution of systems of high-dimensional grid equations during discretization. To improve computing performance, graphics accelerators take a part of the load. To enable this, we developed an algorithm for parallel calculations and implemented it as a software module. As a result of computational experiments, the optimal 2D configuration of streams in a computational unit run on a single streaming multiprocessor was determined. The proposed algorithm and software module enables more efficient use of GPU computational resources when solving computationally intensive hydro physics problems.