半隐式半Lagrangian时间积分及其可扩展并行算法设计

The Scalable Parallel Algorithm Implementation of Semi-Lagrangian and Semi-implicit Time Integration Scheme

目前谱模式仍然是全球数值天气预报业务模式的主流。针对全球数值天气预报谱模式,研究两个时间层的半隐式半Lagrangian时间积分格式以及用于计算起始点的准三次空间插值方法,提出了按需通讯的可扩展并行算法设计,在由4个双CPUSMP结点组成的Linux机群环境下,该算法的8任务相对于4任务的加速比达到了1.65,取得了良好的并行效果。

Spectral models are still the mainstream in global operation NWP now. Based on global spectral model, we discuss twotimelevel semiLagrangian and semiimplicit integration scheme and quasicubic spatial interpolation for the terms evaluated at the departure point. A ondemand massagepassing scalable parallel algorithm has been developed. The algorithm is examined in a 4 nodes dualCPU SMP Linux cluster. The relative speedup of 4 tasks to 8 tasks is 1.65.