Application of the Nvidia CUDA Technology to Solve the System of Ordinary Differential Equations

Abstract

This paper presents some concepts of parallel solution of ordinary differential equations in the context of the Nvidia CUDA technology. Current research leads to the development of algorithms suitable for mass parallel architecture. There has been taken into account the potential opportunities of this architecture, but also significant hardware limitations. Considered conceptions were implemented on both: the Central Processing Unit (CPU) and Graphics Processing Unit (GPU). The results of the computation time measurements and their detailed analysis were provided in this paper.