Acceleration of Boundary-Element Computing Using Graphics Accelerator for the Elements with Nonlinear Form Functions

Abstract

The implementation of a boundary element method (BEM) using three nonlinear form functions designed for determination of half-plane potential distribution was considered in the current paper. The application of NVidia CUDA technology for parallel computing leading to the essential acceleration of computations was also performed. The accuracy of calculations performed with constant and nonlinear form functions was analyzed. The influence of the surface discretization on accuracy was studied. Parallelization technique for BEM using graphic processor was presented. Timings and acceleration coefficient dependencies on the number of boundary elements and calculation points were presented for sequential and parallel calculations.

Keywords

• boundary element method
• acceleration of computing
• potential distribution
• modelling
• cuda
• graphics accelerator
• parallelization