大规模结构有限元分析程序在多核集群计算环境中的性能分析和优化

通过对基于MPI编程模型实现的开源有限元计算分析软件在多核集群计算平台中的程序性能的分析,找出程序瓶颈及其原因,实现了基于MPI编程模型的并行程序在多核计算环境中的性能优化。根据程序性能瓶颈的分析,提出了基于MPI/OpenMP混合并行编程模型的大规模线性/非线性方程组求解和多线程多进程同时进行消息通信的两种程序性能优化方案。不同计算规模的实验结果表明,在多核集群计算平台中,MPI/OpenMP混合编程模型实现的大规模非线性方程组求解器相对于单纯基于MPI编程模型实现的并行程序,其性能有2倍到3倍的提升;多线程多进程同时消息传递的优化方案虽然对程序能够起到性能优化作用,但是对解决程序消息通信瓶颈的问题不是最好的方法。两个方案总体性能分析结果表明,基于MPI/OpenMP混合编程模型实现的并行程序,在多核集群计算平台中能够更好地发挥硬件系统的计算能力。     

并行识别中的依赖关系与通信优化研究

通信开销是阻碍并行程序效率提高的关键因素,降低并行程序的通信开销可以有效地提高程序的运行效率.主要讨论了在并行识别中如何利用依赖关系信息来降低自动并行识别所生成的并行程序的通信开销,包括计算与通信重叠、消除冗余通信代码、消除冗余通信等,并给出了测试结果,试验表明具有较好的优化效果.     

域：支持并行程序概念设计的一种抽象手段

1 引言长期以来,并行计算机在其潜在高性能和现实性能以及简单易用性之间存在着巨大的鸿沟,原因之一在于并行程序设计方法学的滞后导致大量的并行程序设计实践沿袭着传统串行程序开发的思路,而另一个重要的瓶颈在于并行程序的开发缺乏高效编程环境的支持。目前,并行程序的构造主要采取以下两种方式:(一)依据并行算法编写并行程序。这种方式构造的并行程序,一般能达到较高的并行效率,但对广大应用领域的用户要求太高。(二)利用并行化工具对串行程序进行并行改造。由于串行计算机上己经积累了大量成熟的应用程序,通过自动并行编译工具将串行程序移植到并行计算机上运行,无疑具有极为重要的现实意义。并行化方法由于对并行编译的对象缺乏高层的全局信息和并行信息,并行粒度较细,并行效率往往不     

一种并行程序性能工具的研究与实现

随着高性能计算技术的不断发展,并行程序的设计、调试、优化逐渐成为并行技术应用的关键,而性能工具在提高并行程序的执行效率方面发挥着重要的作用。本文阐述了并行程序性能工具的系统结构,以及各个模块功能的原理,并实现了一个基于MPI消息统计的性能工具。     

多核并行编程技术在中文分词程序优化中的应用

为了充分利用多核处理器的硬件资源和计算能力,提出了多核并行编程技术在中文分词程序中的优化方案.根据中文分词最大正向匹配算法的特点,由传统的串行程序,改为并行程序.利用多核并行编程模式的思想,设计了一个混合并行编程模式,通过Intel的性能分析工具,找出了该算法的热点和瓶颈,对其进行优化.实验结果表明,优化过后的执行时间较原来串行程序的执行时间缩短了50%～60%,同时提高了程序的加速性能,取得了良好的效果.     

并行程序概念设计方法的研究

并行程序概念设计方法是将数据并行高层建模语言研究、并行识别方法、并行程序自动构造和人机交互界面技术集成在一起的并行程序设计的一种新方法，能简化并行程序设计，有效缩短并行程序开发周期，提高并行计算效率。文中就上述4个方面的主要技术作了简要介绍，给出了并行程序概念设计系统(PPCDS)的基本框架和计算实例。     

气象资料三维变分同化阶段区域分解并行实现

变分同化由于能明显改善同化质量，正在成为数值天气预报的主流同化方法．研究三维变分同化的并行计算，提出了三维变分同化的阶段区域分解、观测资料的自适应划分算法、计算与通信重叠的矩阵转置和周边区域通信以及文件I／O方法，在此基础上实现了MPI并行三维变分原型系统，在由8个双CPU节点组成的Linux机群上并行加速比达到了11．9．     

OpenMP并行程序的编译器优化

OpemMP标准以其良好的可移植性和易用性被广泛应用于并行程序设计。该文讨论了OpenMP并行程序的编译器优化算法，在编译过程中通过并行区合并和扩展，实现并行区重构，并在并行区中实现了基于跨处理器相关图的barrier同步优化。分析验证表明，这些优化策略减少了并行区和barrier同步的数目，有效地提高了OpenMP程序的并行性能。     

共享内存结构OpenMP并行程序的自动生成

有效的程序自动并行化系统能帮助用户充分利用并行计算机的硬件资源和提高并行程序设计的效率。OpenMP作为共享内存结构的编程标准，具有良好的性能和可移植性。本文介绍了基于SUIF的OpenMP并行程序自动生成工具OAGT的设计和实现，重点讨论了其中所涉及的几个主要技术问题：循环分析、流水并行、归约操作、同步优化等。     

并行可视化调试工具中检查点使能的研究与实现

与串行程序相比，并行程序调试会遇到新的问题。首先并行程序往往需要长时间运行，从而导致并行程序调试是一个尤其费时的过程；其次并行程序调试过程中，某一次调试出现的错误在下次调试的时候不一定出现，给错误跟踪带来了很大困难。本文针对这两个问题，设计和实现了一个中间件系统，在并行调试工具XMPI中使能BLCR检查点系统的。通过该中间件，在使用XMPI调试大型MPI并行程序的时候，减少调试阶段并行程序运行时间，并且可以更好跟踪并行程序错误，提高并行程序开发效率。     

Importance of satellite-retrieved rain/no-rain information on short-range weather predictions

The impact of assimilating rain (satellite-retrieved rainfall is greater than zero) and no-rain (satellite-retrieved rainfall is equal to zero) information retrieved from the Tropical Rainfall Measuring Mission (TRMM) 3B42 precipitation is assessed during Indian summer monsoon 2013 using the weather research and forecasting (WRF) model. Daily three parallel experiments are performed with and without satellite rainfall assimilation for short-range weather forecasts. Additional two experiments are performed daily to evaluate the sensitivity of cumulus parameterization on the WRF model predictions when precipitations are used for assimilation. Precipitation assimilation improves the 48 h low-level temperature, moisture, and winds predictions. Rainfall prediction is also improved over central India when satellite-retrieved rainfall information are assimilated compared to without rainfall assimilation (CNT) experiments. More improvements are seen in moisture forecasts when the Kain–Fritsch (KF) cumulus convection parameterization scheme is used against the Grell–Devenyi ensemble (GD) scheme, whereas for temperature and wind speed forecasts the Grell convection parameterization scheme performed better over the Indian region. Overall, precipitation assimilation improved the WRF model analysis and subsequent model forecasts compared with without precipitation assimilation experiments. Results show that no-rain observations also have a significant positive impact on short-range weather forecasts.     

Evaluation and assimilation of various satellite-derived rainfall products over India

Accurate prediction of rainfall from the numerical weather prediction model is one of the major objectives over tropical regions. In this study, four different satellite-derived rainfall products (viz. merged-rainfall product from TRMM (Tropical Rainfall Measuring Mission) 3B42 and IMERG (Integrated Multi-satellitE Retrievals for GPM (Global Precipitation Measurement)), and Indian meteorological satellite INSAT-3D retrieved HEM (Hydro-Estimator Method) and IMSRA (INSAT Multi-Spectral Rainfall Algorithm) rainfall) are assimilated in the Weather Research and Forecasting (WRF) model using variational method. Before assimilation of satellite retrieved rainfall product in the WRF model, selected rainfall products are compared with ground rainfall from India Meteorological Department during Indian summer monsoon (June–September) 2015. Preliminary validation results show root-mean-square-difference (mean difference) of 18.1 (2.1), 21.3 (2.1), 15.4 (?0.72), and 14.4 (0.5) mm day^?1 in IMSRA, HEM, IMERG, and TRMM 3B42 rainfall, respectively. Further, the four-dimensional variational data assimilation method is used daily to assimilate selected rainfall products in the WRF model during the entire month of August 2015. Results suggest that assimilation of satellite rainfall improved the WRF model analyses and subsequent temperature and moisture forecasts. Moreover, rainfall prediction is also improved with the maximum positive impact from TRMM rainfall assimilation followed by IMERG rainfall assimilation. Similar nature of improvements is also seen in rainfall prediction when INSAT-3D retrieved rainfall products (HEM and IMSRA) are used for assimilation.     

Sensitivity analysis of high-resolution Oceansat-2 scatterometer winds on Thane cyclone simulation

Ocean surface wind vectors retrieved from the Oceansat-2 scatterometer (OSCAT) are used in this study to evaluate their impact on Thane cyclone simulation. The Weather Research and Forecasting (WRF) model and its three-dimensional variational (3D-Var) data assimilation system are adapted to evaluate the sensitivity of OSCAT observations. Simulated track error and landfall forecast are considered as standard measurements to assess the impact of 50 km and ~15 km spacing grid OSCAT winds along and across the swath. Significant improvement is obtained in track forecasting, when high-resolution vector winds (HVW; composite slice-level winds, ~15 km) are used for assimilation rather than coarser-resolution (50 km) operational OSCAT winds. Forecasting sensitivity to observations (OSCAT winds) using WRF tangent linear and adjoint modelling is used to quantify the impact of two different resolutions of OSCAT winds. WRF adjoint modelling is used here as a diagnostic tool, which indicates that high-resolution OSCAT winds have a more positive impact on the track prediction of Thane tropical cyclone.     

The impact of background error parameters on heavy rainfall prediction

A heavy rainfall event over the northwest of India is selected to investigate the impact of Atmospheric Infrared Sounder (AIRS)-retrieved temperature and moisture profile assimilation on regional model prediction. The Weather Research and Forecasting (WRF) model and its three-dimensional variational (3D-Var) data assimilation system (WRFDA) is used to assimilate AIRS profiles with tuning of two major background error parameters – viz. length and variance scales. Assimilation of AIRS profiles improves the WRF model analyses, which are closer to the Moderate Resolution Imaging Spectrometer (MODIS) profiles compared to those without assimilation experiment. Results show that within a wide parameter range of length and variance scales, the assimilation of AIRS-retrieved profiles has a positive influence on heavy rainfall prediction. Approximately 9–30, 5–42, and 0.5–3.0% domain average values of improvement are observed after AIRS profile assimilation for different values of length and variance scales in temperature, water vapour mixing ratio, and rainfall prediction, respectively. This study shows that the impact of observations on the WRF model forecast is dependent on the length and variance scale parameters of background error, and lower values of length scale in WRFDA result in degradation of the forecast.     

同化FY-3C/MWHTS观测资料反演的海面气压场对台风数值预报的影响

由中国风云三号C星（FY-3C）搭载的微波温湿探测仪（MWHTS）的亮温观测资料能够实时反演得到高分辨率、高精度的海面气压场。基于三维变分同化方法将FY-3C/MWHTS观测资料反演的海面气压场同化进入中尺度天气研究与预报（Weather Research and Forecasting, WRF）模式,以台风“Maria”和“Noru”为例,通过控制实验和同化试验的对比分析,探讨了同化反演的海面气压场对台风数值预报的影响。初始化敏感性试验结果表明,同化海面气压场使初始时刻台风中心气压与位置更接近实况,并且调整了台风初始温度场和风场的结构和分布。台风的数值预报结果表明:同化反演的海面气压场能够改进台风的路径和强度预报精度。     

Influence of Assimilating the Sea Surface Pressure Fields Retrieved from FY-3C/MWHTS Data on Typhoon Forecasting

Using measurements with the Microwave Temperature and Humidity Sounder (MWHTS) onboard the Chinese Fengyun-3C satellite, real-time and high resolution sea surface pressure information can be retrived. Based on the three-dimensional variational assimilation (3DVAR) method, the retrieved sea pressure fields from FY-3C/MWHTS observations are assimilated into the Weather Research and Forecasting (WRF) model. The influence of the retrieved pressure fields on typhoon forecasting is discussed through the comparison between control experiment and assimilation experiment. Sensitivity experiments of typhoon Maria and Noru show that the assimilation of sea surface pressure fields makes the central pressure and central location closer to the actual value, and adjusts the structure and distribution of initial temperature fields and wind fields. The numerical prediction results show that the assimilation of the sea surface pressure fields can improve the accuracy of typhoon track and intensity prediction.     

A comparative study of libRadtran and RTTOV predicted satellite images using the WRF forecasting output

The assimilation of cloudy radiances remains important in improving precipitation and severe weather forecasting. In practice, Numerical Weather Prediction (NWP) Models frequently do not predict meso-scale phenomena, so the phenomenon is either predicted but not realised, or is well predicted but not where it is observed. Radiative Transfer Models such as TIROS-Television and Infrared Observation Satellite Operational Vertical Sounder (RTTOV) and libRadtran are the mathematical operators used in the simulation of satellite data. In the data assimilation process, an effective reproduction of the mesoscale convective phenomenon leads to high quality data analysis. Therefore, we are looking for an operator that reproduces the NWP model’s behaviour in a realistic way. Several cloud parameterisation schemes are available in RTTOV and libRadtran to simulate the satellite cloudy radiances. Each selected scheme may result in different simulated brightness temperature data compared to those observed by satellite. However, the source of errors is still unknown: are they generated by the RT model, are they coming from the predicted fields of the NWP Model used as input, or from both? This study aims to investigate the impact of libRadtran or RTTOV operators on the quality of the predicted satellite image. The same mandatory forecasted fields are used as input for both models and derive from the Weather Research and Forecasting (WRF) Limited-Area Model. In this study, we did not investigate the total capability of RT models, but we have focused on a standard and specific physical parameterisation scheme. As reference data, Meteosat Second Generation (MSG) images have been used to compute the deterministic and probabilistic scores. The results for the deterministic scores show that libRadtran reproduces the cold temperatures predicted by WRF well, but these are sometimes slightly distant from their geographical location. Conversely, for the RTTOV, there is a tendency to miss more cases of good detection of the events predicted by WRF. Probabilistic analyses confirm an improvement in libRadtran scores when the neighbourhood size is increased, and a boxplot analysis of the bootstrap method confirms the stability of scores for both models.     

Optimal design of a 3-PUPU parallel robot with compliant hinges for micromanipulation in a cubic workspace

In recent years, nanotechnology has been developing rapidly due to its potential applications in various fields that new materials and products are produced. In this paper, a novel macro/micro 3-DOF parallel platform is proposed for micro positioning applications. The kinematics model of the dual parallel mechanism system is established by the stiffness model with individual wide-range flexure hinge and the vector-loop equation. The inverse solutions and parasitic rotations of the moving platform are obtained and analyzed, which are based on a parallel mechanism with real parameters. The reachable and usable workspace of the macro motion and micro motion of the mechanism are plotted and analyzed. Finally, based on the analysis of parasitic rotations and usable workspace of micro motion, an optimization for the parallel manipulator is presented. The investigations of this paper will provide suggestions to improve the structure and control algorithm optimization for the dual parallel mechanism in order to achieve the features of both larger workspace and higher motion precision.     

The impact of assimilating MeghaTropiques SAPHIR radiances in the simulation of tropical cyclones over the Bay of Bengal using the WRF model

The present study aims to investigate the impact of assimilating SAPHIR (Sounder for Probing Vertical Profiles of Humidity) radiances in the simulation of tropical cyclones over the Indian region by the Weather Research and Forecasting (WRF) model. Three tropical cyclones which formed over the Bay of Bengal are chosen as the case studies. Since SAPHIR is a humidity microwave sensor, it is interesting to assess the impact of these observations in simulating cyclones which depend significantly on moist-convective processes. The study makes use of the three-dimensional variational (3DVar) assimilation technique of the WRF variational assimilation system. The results of the study indicate that the assimilation of SAPHIR radiances do have a positive impact on the simulation of tropical cyclones considered here. Two model simulations are performed – a control run (Ctrl) with only conventional and satellite wind observations assimilated, and a SAPH run (SAPH) where SAPHIR radiances are also assimilated in addition to conventional and satellite wind observations. Both these simulations are compared to each other and to observations from the India Meteorological Department (IMD), Joint Typhoon Warning Centre (JTWC), and Tropical Rainfall Measurement Mission (TRMM), as well as analysis fields from Global Forecast System (GFS) from the National Centres for Environmental Prediction (NCEP). Comparison of minimum sea level pressure and maximum wind speed simulated by the model with the IMD and JTWC observations shows that the SAPHIR assimilation has a moderate impact on the simulation of these features by the model. Track prediction of the model is also improved at initial forecast times, as evidenced by the reduced track errors in the model run with SAPHIR radiances assimilated. The warm core structure, as well as the relative vorticity structure of the cyclones, are also impacted in a moderate manner by the assimilation of SAPHIR radiances. The assimilation also positively impacted the rainfall simulation of the model. This is seen from the higher equitable threat score, lower false alarm ratio, and higher probability of detection estimated with respect to TRMM observations, in the SAPH run as compared to the Ctrl run.     

面向OpenMP和OpenTM应用的并行数据重用理论

将经典的数据重用理论扩充到并行领域,分别提出了面向OpenMP和OpenTM应用的并行数据重用理论.针对重用在线程、事务中的关系,系统地讨论了并行应用中重用的分类、判定和求解方法.同时,应用这一理论研究了OpenTM循环的优化技术,以降低事务被回退的风险.最后,使用并行数据重用理论分析和统计了SPEComp2001中的数据重用.并行数据重用理论可以用于指导面向多核存储共享结构的并行程序分析和编译优化技术研究.