GPU driven finite difference WENO scheme for real time solution of the shallow water equations - ScienceDirect
In recent years, with the continuous improvement of computing power, some parallel computing techniques with different programming languages have been proposed, such as the multi-CPUs parallel and CPU-GPU heterogeneous parallel computing techniques based on MPI and CUDA software. Moreover, it is easier to implement the parallel techniques for the meshless method with local matrix or particle method over the conventional grid-based methods with global matrix system [84,85]. For example, the SPH particle methods with multi-CPUs or GPU-enabled parallel computing [53,86-89] are widely used and the Eulerian WLS methods with GPU-enabled parallel computing [90,91] have been effectively proposed to solve the general PDE or complex nonlinear PDEs. It is also clear and well-established that a certain set of properties must be ensured by the numerical scheme for the solution to be accurate and efficient under complex and realistic flow conditions (Bermudez and Vázquez, 1994; Caleffi and Valiani,