Parallelization of the distinct lattice spring model

The distinct lattice spring model (DLSM) is a newly developed numerical tool for modeling rock dynamics problems, i.e. dynamic failure and wave propagation. In this paper, parallelization of DLSM is presented. With the development of parallel computing technologies in both hardware and software, parallelization of a code is becoming easier than before. There are many available choices now. In this paper, Open Multi‐Processing (OpenMP) with multicore personal computer (PC) and message passing interface (MPI) with cluster are selected as the environments to parallelize DLSM. Performances of these parallel DLSM codes are tested on different computers. It is found that the parallel DLSM code with OpenMP can reach a maximum speed‐up of 4.68× on a quad‐core PC. The parallel DLSM code with MPI can achieve a speed‐up of 40.886× when 256 CPUs are used on a cluster. At the end of this paper, a high‐resolution model with four million particles, which is too big to handle by the serial code, is simulated by using the parallel DLSM code on a cluster. It is concluded that the parallelization of DLSM is successful. Copyright © 2011 John Wiley & Sons, Ltd.     

MPI环境下基于边界校正方法的海洋模型的并行化

基于区域分解技术,设计了一种针对海洋模型的边界校正方法,该方法不需要对原有海洋模型进行较大改动,只需要利用消息传递接口(message passing interface,MPI)编写子区域间的数据传输模块,即可实现原有海洋模型的并行化。相对于重新开发并行系统,该方法能够较快把原有模型的串行解决方案转化为并行解决方案,并且并行化后的海洋模型能够获得较高的加速比。     

基于移动多核GPU的并行二维DCT变换实现方法

传统的基于CPU的串行程序所实现的二维DCT变换算法时间复杂度高变换效率低,难以满足许多应用的实时要求。特别是在当代以嵌入式处理器为核心的移动端信息处理终端,有限的CPU性能更加难以实现快速的DCT变换。值得欣慰的是新一代嵌入式处理器提供了支持GPGPU技术的GPU,为解决复杂的移动计算问题提供了高效的并行化解决途径。基于最新的ARM Cortex-A15内嵌GPU Mali-T604及Open CL框架设计实现了一种针对二维DCT变换的并行化加速方案并实测了优化效果,实验结果表明文中的并行方案能够提高二维DCT变换的效率,在输入数据量足够大的条件下能够达到近20倍的加速比。     

Coupled 2D Hydrodynamic and Sediment Transport Modeling of Megaflood due to Glacier Dam-break in Altai Mountains,Southern Siberia

One of the largest known megafloods on earth resulted from a glacier dam-break,which occurred during the Late Quaternary in the Altai Mountains in Southern Siberia.Computational modeling is one of the viable approaches to enhancing the understanding of the flood events.The computational domain of this flood is over 9460 km2 and about 3.784 × 106 cells are involved as a 50 m × 50 m mesh is used,which necessitates a computationally efficient model.Here the Open MP(Open Multiprocessing) technique is adopted to parallelize the code of a coupled 2D hydrodynamic and sediment transport model.It is shown that the computational efficiency is enhanced by over 80% due to the parallelization.The floods over both fixed and mobile beds are well reproduced with specified discharge hydrographs at the dam site.Qualitatively,backwater effects during the flood are resolved at the bifurcation between the Chuja and Katun rivers.Quantitatively,the computed maximum stage and thalweg are physically consistent with the field data of the bars and deposits.The effects of sediment transport and morphological evolution on the flood are considerable.Sensitivity analyses indicate that the impact of the peak discharge is significant,whilst those of the Manningroughness,medium sediment size and shape of the inlet discharge hydrograph are marginal.     

Control volume method for hydromagnetic dynamos in rotating spherical shells: Testing the code against the numerical dynamo benchmark

Hydromagnetic dynamos in rotating spherical shells are investigated using the control volume method. We present a validation of our code against the numerical dynamo benchmark. It is successfully benchmarked and we are able to conclude that the control volume method is another numerical method available for numerical modelling of self-consistent dynamos. In addition, the efficiency of our numerical code is tested. Computations provide conclusions that dynamo codes based on the spectral methods are much more efficient than our code based on the control volume method at the study of global fields on small and medium size parallel computers. However, our code could be much more efficient than codes based on the spectral methods on very large parallel computers, especially at the study of turbulence.     

中期数值预报系统T213L31在IBM/SP高性能计算机上的建立

在引进欧洲中期天气预报中心 (ECMWF) 的全球谱模式的基础上，通过对原模式的分析改造，首次以分布与共享相结合的方式在国家气象中心IBM/SP高性能计算机上实现了全球谱模式的高效运行。采用调整向量长度、优化程序设计、完善消息传递机制和实现MPI与OpenMP的混合并行编程等方法，减少模式的通信量、计算量和内存的使用量，提高了计算效率。实现了在T213L31分辨率条件下，10天预报可以在3 h之内完成，达到业务对时限的要求。建立了与T213L31全球谱模式相配套的最优插值（OI）并行处理分析系统，解决了由于观测站点在全球不均匀分布所带来的计算负载不均衡问题。在此基础上，实现了T213L31全球资料同化与预报系统并建立了相应的自动作业监控系统。     

多重网格区域分裂分布式计算

对分布式多重网格计算进行了研究.其顺序算法描述的是非递归形式, 算法并行化是基于区域分裂实现的.网状拓扑结构组织在多处理机上, 并行算法映射到多进程上, 在一定程度上显著提高并行化速度和并行化效率      

Parallel optimal choropleth map classification in PySAL

In this article, we report on our experiences with refactoring a spatial analysis library to support parallelization. Python Spatial Analysis Library (PySAL) is a library of spatial analytical functions written in the open-source language, Python. As part of a larger scale effort toward developing cyberinfrastructure of GIScience, we examine the particular case of choropleth map classification through alternative parallel implementations of the Fisher-Jenks optimal classification method using a multi-core, single desktop environment. The implementations rely on three different parallel Python libraries, PyOpenCL, Parallel Python, (PP) and Multiprocessing. Our results point to the dominance of the CPU-based Parallel Python and Multiprocessing implementations over the Graphical Processing Unit (GPU)-based PyOpenCL approach.     

基于移动多核GPU的并行二维DCT变换实现方法

传统的基于CPU的串行程序所实现的二维DCT变换算法时间复杂度高变换效率低,难以满足许多应用的实时要求.特别是在当代以嵌入式处理器为核心的移动端信息处理终端,有限的CPU性能更加难以实现快速的DCT变换.值得欣慰的是新一代嵌入式处理器提供了支持GPGPU技术的GPU,为解决复杂的移动计算问题提供了高效的并行化解决途径.基于最新的ARM Cortex-A15内嵌GPU Mali-T604及OpenCL框架设计实现了一种针对二维DCT变换的并行化加速方案并实测了优化效果,实验结果表明文中的并行方案能够提高二维DCT变换的效率,在输人数据量足够大的条件下能够达到近20倍的加速比.     

大气环流数值模式的一种并行化方案

从一般计算机并行计算理论出发,结合大气科学模式的特点,总结出一套对数值模式适用的并行化编程方案,而且提出了并行计算过程中应该注意的一些问题.利用文中提到的方案,可以帮助刚开始接触并行计算的模式编程人员快速完成串行数值模式的并行编程.最后,以一个串行的大气环流模式SAMIL并行化编程作为实例,对其并行计算过程和并行计算效率进行了描述和分析.