集群式高性能计算系统研究

研究了集群的系统结构和主要优势,以及集群式高性能计算系统的诞生;分析了集群式高性能计算系统的架构和构建方式,集群构建包括网络部署、存储系统、计算节点、管理节点、登录节点等部分。在此基础上构建了基于Linux的集群式高性能计算系统。     

基于高性能计算的开源云平台性能评估

云计算是一种提供各种IT服务的互联网资源利用的新模式,已经广泛地应用在包括高性能计算的各种领域。然而,虚拟化带来了一些性能开销;同时,不同的云平台实施虚拟化技术的不同,使得在这些云平台上应用高性能计算服务的性能也千差万别。通过HPC Challenge (HPCC) Benchmark和NAS Parallel Benchmark(NPB)分别对CPU、内存、网络、扩展性和高性能计算真实负载进行评估,比较并分析了诸如Nimbus、OpenNebula和OpenStack实施高性能计算的性能,实验显示OpenStack对计算密集型的高性能应用负载表现出较好的性能,因此,OpenStack是实施高性能计算的开源云平台的一个好的选择。     

An evaluation of PCs for high energy physics under Windows NT and Linux

The high availability and commodity prices of Intel-based PCs have made them serious contenders to workstations. We address the question whether PCs are appropriate for running real physics codes. We evaluate the performance of Intel-based systems with two popular 32-bit operating systems, Linux and Windows NT. We report on a suite of benchmark tests, both generic and HEP-specific, where we compare the performance of these two systems to each other, and to other popular workstation class machines.     

基于局域网和MPI的PC集群计算环境

利用现成PC构建由几十乃至几百台PC组成的廉价、实用且性能优良的并行计算机。实验系统是在由40台PC组成的以太局域网内,建立一个基于TRUBO—LINUX和MPI的集群计算环境,并在此基础上进行并行计算实验和性能测试。实验表明,这种环境适合于那些进程之间通信不频繁的或者通信开销相对于计算开销小得多的中粒度或粗粒度的计算任务。     

用于并行计算的PC集群系统构建*

在注射成形模拟研究过程中,涉及材料的牛顿和非牛顿黏性流动模拟和注射成形后期的冷却过程模拟,以及随时间变化各处的压力变化等科学和工程领域经常应用大规模科学计算。随着基于网格的计算和数据处理日益复杂,很多计算一般PC系统无法满足要求,需要超级计算环境。因为不断追求更高的计算精度和日益复杂的对象而扩大计算规模,传统的串行处理方式难以满足这些要求。因此,现代高性能计算的低成本、高效率成为选择并行计算的解决方式。重点阐述如何构建一个用于并行计算的PC集群系统,结合实例阐明MPI的实现方法,以及对PC集群系统进行了性     

Windows NT仍将是工业PC理想的工作平台

本文就WindowsNT操作系统所希望达到的目标、结构体系以及网络运用等方面进行了较为详细的介绍，同时又将其与其它比较流行的UNIX、OS／2、Windows3．X等操作系统进行比较对比，阐明了对于工业自动化系统的应用特性，提出了WindowsNT仍将是工业PC理想工作平台的观点。     

基于集群技术构建聚变研究高性能计算系统

高性能计算是解决大规模聚变模拟计算的有效途径.介绍如何基于集群技术构建聚变研究高性能计算系统,分析了系统的基本体系结构,从硬件和软件两个方面详细解析其高性价比的构建技术.最后,运用基准测试程序(LinPACK和NPB) 对这一系统进行测试,显示了该系统高效的并行计算性能.     

基于Windows95／NT的PVM并行计算平台

讨论在Ｗｉｎｄｏｗｓ下实现的并行虚拟计算机（ＰＶＭ ｆｏｒ Ｗｉｎｄｏｗｓ）的若干技术。     

对Windows NT核心的实时性能的分析与测试

文章分析了Windows NT核心的体系结构和其实时性能之间的关系,设计了测试NT内核实时响应能力的实验,并得到了一些有关NT内核的重要数据.     

Comparing Windows NT,Linux, and QNX as the basis for cluster systems

Clusters use commodity hardware and software components to provide an environment for high-performance parallel processing. A major issue in the development of a cluster system is the choice of the operating system that will run on each node. We compare three alternatives: Windows NT, Linux, and QNX—a real-time microkernel. The comparison is based on expressive power, performance, and ease-of-use metrics. The result is that none of these systems has a clear advantage over the others in all the metrics, but that each has its strong and weak points. Thus any choice of a base system will involve some technical compromises, but not major ones. Copyright © 2001 John Wiley & Sons, Ltd.     

Windows NT下对磁盘性能监测的研究

基于Windows NT(5.0以上)操作系统,通过多种方式(应用层、内核层)实现Windows NT下对物理磁盘的读写操作性能的监测,用以实现对传统机械磁盘以及SSD性能的改进研究.介绍了ETW (Windows事件追踪)机制的基本架构、原理,及ETW技术在物理磁盘读写操作性能监测方面的具体实现;讨论了WMI在磁盘性能监测方面的应用;在Windows NT下对数据读写从应用层到驱动层的路径进行分析,介绍了内核中驱动HOOK技术的原理及其特点;提出驱动层HOOK技术在物理磁盘读写操作性能监测方面的具体实现.最后介绍了驱动方式实现物理磁盘读写操作性能监测的应用前景.     

Windows NT设备驱动程序设计

介绍了Ｗｉｎｄｏｗｓ ＮＴ,Ｗｉｎｄｏｗｓ核心驱动程序的一般体系结构及Ｗｉｎｄｏｗｓ ＮＴ驱动程序的设计方法,实现了ＩＳＡ接口数据采集卡从Ｗｉｎｄｏｗｓ９５平台到Ｗｉｎ－ｄｏｗｓ ＮＴ平台的转移。     

Planning for Windows NT 5.0 security

Microsoft has announced that Windows NT 5.0 will be released in December 1998. This version of Windows NT will be different from all previous releases of this product in a number of very significant ways. This article provides an initial examination of the expected security-related features, functionality, and security exposures of Windows NT 5.0 to assist in planning for and providing needed levels of security control. planning for and providing needed     

AMRC: an algebraic model for reconfiguration of high performance cluster computing systems at runtime

High Performance Cluster Computing Systems (HPCSs) represent the best performance because their configuration is customized regarding the features of the problem to be solved at design time. Therefore, if the problem has static nature and features, the best customized configuration can be done. New generations of scientific and industrial problems usually have dynamic nature and behavior. A drawback of this dynamicity is that the customized HPCSs face challenges at runtime, and consequently show the worse performance. The reason for this might be due to the fact that dynamic problems are not adapted to configuration of the HPCS. Hence, requests of the dynamic problem are not in the direction of the HPCS configuration. The main proposed solutions for this challenge are dynamic load balancing or using reconfigurable platforms. In this paper, a vector algebra-based model for HPCS reconfiguration at runtime is presented and named AMRC. This model determines the element causing the dynamic behavior and analyzes the reason regarding both software and hardware at runtime. Some results of the presented model show that by defining a general state vector whose direction is toward reaching high performance computing and whose weight is based on the initial features and explicit requirements of the problem, as well as by defining a vector for each process in the problem at runtime, we can trace changes in the directions and uncover the reason for them.     

基于Windows CE&Pocket PC的数据库应用程序开发

阐述了WINCE&P/PC应用程序设计的特点,讨论了运 用ADOCE开发数据库应用程序的步骤及关键点,最后给出 了发布应用程序的方法。     

基于高性能计算的SWAT参数敏感度分析并行框架

随着大规模水文模拟需求的不断提高,如何解决计算需求问题逐渐成为水文研究的一个热点.SWAT(soil and water assessment tool)模型在进行大规模水文模拟时有着良好的适应性与准确度,但其敏感度分析模块由于计算量过高,计算时长往往长达数月之久.为了加快SWAT敏感度分析的运行速度,针对SWAT敏感度分析模块的特点,基于MPI提出了一种高效的主—从式并行计算框架,并在此框架的基础上,通过将正演过程并行化,在敏感度分析的主—从并行框架中引入通信子空间的操作,将并行化的正演与主—从式的外层并行框架相结合,得到一种混合式的敏感度分析并行框架,大大提高了对参数集合的敏感度分析速度,将SWAT敏感度分析模块使用的处理器数量从原始的单核串行一跃提升到百核的数量级.最后通过天山北坡流域的模拟验证了此并行框架的可行性.     

基于异步光分组交换的高性能计算机系统

基于异步光分组交换及光组播技术,提出一种新型高性能计算机系统（HPCS）。采用分布式控制结构进行系统的分级管理,有利于处理大批量突发业务及系统的全光多级扩展。基于循环光纤延时线（Rec-FDL）,提出一种紧凑、大容量的异步全光分组冲突解决机制,并与传统的分组重传冲突解决算法（PRA）进行了性能比较;建立了系统的稳定性、丢包率及时延分析模型。仿真结果表明,对于一个能够实现40 320个CPU互连的两级系统,Rec-FDL机制的平均分组等待时延TRec-FDL约为0.004 ns,而PRA算法的平均分组等待时延Trerans随分组传输距离D增加而增加,在D为10 m和20 m时,Trerans分别等于0.006 ns和0.01 ns。因此,对于分布式HPCS而言,Rec-FDL机制具有更优秀的时延特性。