The Research Of Parallel Algorithm Based On CUDA For Hydrological Model

With the rapid development of computer technology, remote sensing, GIS, such as access to a wide range of spatial data and information for hydrological model provides a convenient, especially the development of geographic information systems, but also for the hydrological model better, more convenient use of space data provide a good platform. Despite the large number of spatial data acquisition makes hydrological model simulations for hydrological processes more scientific, more accurate, but also greatly increases the amount of computation of hydrological models, hydrological model allows simulation speed slows down, the efficiency is low. As a CUDA parallel computing architecture that enables multi-threaded parallel manner, greatly improving the speed of operation of a variety of operations. Near to CUDA has been widely used in various fields in order to improve the efficiency of research and production. Therefore, this article using CUDA technology to improve the efficiency of hydrological simulation models studied, hope through research,and find out CUDA-based technologies to improve the hydrological model simulation efficiency.This paper based on the different types of hydrological models or hydrologic model of a technology-based CUDA parallel algorithm rewritten, and use the method of comparative analysis the conventional algorithms; the parallel algorithms based on CUDA computing efficiency; determined based on the CUDA parallel computing methods and procedures for hydrological model based on the results; suitable for parallel computing for hydrological model, and what factors affect the degree of operational efficiency.By raster-based spatial analysis, water extraction and Xinanjiang model research model to determine the impact of hydrological model based on CUDA parallel computing efficiency factors are the following:1) The amount of data parallel computingTo CUDA parallel computing based on the host side and device-side data transmission, data size will affect the efficiency to some extent.2) Computational complexityThe higher complexity of the operation, the longer a single operation, efficiency more obvious after the parallel computing.3) The parallelizable degree of hydrological modelCan influence the model’s parallel degree has the following main factors: First, the space parallel; second sub-process can parallelism; the third is the time to parallelism.For the hydrological model, the sub-process can parallelism although it can enhance operational efficiency, but in the overall structure of the serial case, the promotion is limited. Because of hydrological processes in real time in close contact often, so the time may be very limited parallelism. Therefore parallelism space is the main factor affecting the degree of parallelism of the hydrological model.