论文部分内容阅读
本文基于GPU(Graphic processor unit)并行算法建立了非结构网格有限体积法海洋模型(GPUOM),并实现了对长山群岛的潮流数值模拟。计算结果表明,该模型可以对岛群复杂岸线及地形条件下的潮流运动进行精细准确的计算。此外,通过对比GPU-OM、FVCOM和MIKE 21的并行加速能力,本文模型在Tesla K20X上的计算效率明显优于32线程的FVCOM模型和48线程的MIKE 21并行计算模式。
In this paper, an unstructured grid finite volume ocean model (GPUOM) is established based on parallel algorithm of GPU (Graphic processor unit), and the tidal current numerical simulation of Changshan archipelago is realized. The calculation results show that the model can accurately calculate the tidal current under complex island shoreline and topography. In addition, by comparing the parallel acceleration capabilities of GPU-OM, FVCOM, and MIKE 21, the computational efficiency of this model on the Tesla K20X is significantly better than the 32-thread FVCOM model and the 48-thread MIKE 21 parallel computing model.