摘要
为了提高河口水沙数学模型的计算速度,通过基于共享内存的OpenMP和基于区域分解的消息传递模型(MPI并行)两种并行方法,将串行的河口水沙数学模型并行化。将并行模型应用到钱塘江河口潮流泥沙运动模拟实例中。结果表明,该并行模型能大大提高计算速度。其中,OpenMP方法并行效率较低,而基于区域分解的MPI并行有很好的加速效果,在64个处理器上加速比可接近36,具有较大的推广价值。
In order to improve the computing speed of flow and sediment mathematic model in estuaries,the series flow and sediment mathematical model is parallelized by memory shared OpenMP and information transmission MPI based on domain decomposition.The parallel model is applied in a practical flow and sediment movement simulation in Qiantang River in China.The result shows that the parallel model can improve the computing speed greatly;however,the OpenMP method has lower parallel efficiency,MPI method on the basis of domain decomposition is effective in accelerating the computing speed,the speedup ratio by 64 processors reaches 36.
出处
《人民长江》
北大核心
2013年第15期54-57,共4页
Yangtze River
基金
国家自然科学基金资助项目(40806037)
水利部公益性行业专项(201101056)
关键词
水沙数学模型
区域分解
并行化
MPI
Metis
OPENMP
河口
flow and sediment mathematic model
domain decomposition
parallel computer technology
MPI
Metis
OpenMP
estuary.
