中国科学技术大学：《并行计算 Parallel Computing》课程教学资源（习题）第二章 当代并行计算机系统介绍

第二章当代并行计算机系统介绍 习题例题: 1.请尽可能访问以下有关高性能并行计算的网址: IeeE/cs Parascope(htTp: //computer. org/parascope/), world-wide parallel computing sites High Performance Computing Lists (http://www.cs.colorado.edw/homes/mcbryan/publichtml/bb/2/summary.htmi) TheLanguageList(http://cuiwww.unige.ch/langlist)enumerateprogramminglanguages Top500(http://www.netlib.org/benchmark/top500.html) Worlds TOP 500 most powerful computing sites(at Netlib, University of Tennessee Myrinet(http://www.myri.com) Dsmbibliography(http://www.cs.ualberta.ca/-rasit/dsmbiblio.html BerkeleyActiveMessagepage(http://now.cs.berkeleyedw/am/activemessageshtml TheCrayResearchsystempage(http://www.cray.com/products/systems/) Sgl/CrayOrigin2000(http://www.sgi.com/products/haRdware/servers/index.html CrayT3e(htTp://www.cray.com/products/systems/crayt3e/) Petaflopswebsite(http://www.aero.hq.nasagov/hpcc/) NasaHpccProgram(http://www.aero.hq.nasagov/hpcc/) CrayT3e(httP://www.cray.com/products/systems/crayt3e/) IbmSp(hTtp://www.rs6000.ibm.com/hardware/largescal IntelParagon(http://www.sdsc.edu/serviceS/consult/paragon/paragon.html) KaiLi(http://www.cs.princetonedu/-l Sp2atMhpcc(http://www.mhpcc.edu/doc/sp2.general/sp2.generalhtml) MpiStandardsite(http://www.mcs.anl.gov/mpi/index.html MitParallelandDistributedOperatingSystemsGroup(http://www.pdos.Ics.mitedu/ National Center for Supercomputer Applications at UIUC (NCSA) (http://www.ncsa.uiuc.edu/) CornellTheoryCenter(ctc)(http://www.tc.corneledu/ctc.html ArgonneNatlLaboratory,Mathematics&ComputerScienceDiv.(http://www.mcs.anl.gov/) ArmyResearchLab(http://www.arl.armymin) awrenceLivermoreNationalLaboratory(http://www.IiNl.g oV/comp/comp Los Alamos Natl Laboratory(LANL) Advanced Computing Laboratory MauiHighPerformanceComputingCenter(mhpcc)(http://www.mhpcc.edu/mhpcc.html) SanDiegoSupercomputerCenter(http://www.sdsc.edu/sdschomE.html) SandiaNationalLaboratories(http://www.cs.sandiagov/)MassivelyParallelComp.Res ParallelProcessinginJapan(http://fuji.stanford.edu/papers/ppij.html CrayResearch(http://www.cray.com/) IbmHigh-performanceComputing(http://ibm.tc.cornelledu/) ParasoftCorporation(http://www.parasoft.com/)

第二章 当代并行计算机系统介绍 习题例题： 1. 请尽可能访问以下有关高性能并行计算的网址： IEEE/CS ParaScope (http://computer.org/parascope/)，world-wide parallel computing sites High Performance Computing Lists (http://www.cs.colorado.edu/homes/mcbryan/public_html/bb/2/summary.html) The Language List (http://cuiwww.unige.ch/langlist) enumerate programming languages TOP 500 (http://www.netlib.org/benchmark/top500.html) World's TOP 500 most powerful computing sites (at Netlib，University of Tennessee) Myrinet (http://www.myri.com) DSM bibliography (http://www.cs.ualberta.ca/~rasit/dsmbiblio.html) Berkeley Active Message page (http://now.cs.berkeley.edu/AM/active_messages.html) The Cray Research system page (http://www.cray.com/products/systems/) SGI/Cray Origin 2000 (http://www.sgi.com/Products/hardware/servers/index.html) Cray T3E (http://www.cray.com/products/systems/crayt3e/) PetaFLOPS web site (http://www.aero.hq.nasa.gov/hpcc/) NASA HPCC Program (http://www.aero.hq.nasa.gov/hpcc/) Cray T3E (http://www.cray.com/products/systems/crayt3e/) IBM SP (http://www.rs6000.ibm.com/hardware/largescale/) Intel Paragon (http://www.sdsc.edu/Services/ Consult/Paragon/paragon.html) Kai Li (http://www.cs.princeton.edu/~li/) SP2 at MHPCC (http://www.mhpcc.edu/doc/SP2.general/SP2.general.html) MPI Standard site (http://www.mcs.anl.gov/mpi/index.html) MIT Parallel and Distributed Operating Systems Group (http://www.pdos.lcs.mit.edu/). National Center for Supercomputer Applications at UIUC (NCSA) (http://www.ncsa.uiuc.edu/) Cornell Theory Center (CTC) (http://www.tc.cornel.edu/ctc.html) Argonne Natl Laboratory，Mathematics & Computer Science Div. (http://www.mcs.anl.gov/) Army Research Lab (http://www.arl.army.mil/) Lawrence Livermore National Laboratory (http://www.llnl.gov/comp/comp.html) Los Alamos Natl Laboratory (LANL) Advanced Computing Laboratory (http://www.acl.lanl.gov/). Maui High Performance Computing Center (MHPCC) (http://www.mhpcc.edu/mhpcc.html) San Diego Supercomputer Center (http://www.sdsc.edu/SDSCHome.html) Sandia National Laboratories (http://www.cs.sandia.gov/) Massively Parallel Comp. Res. Lab. Parallel Processing in Japan (http://fuji.stanford.edu/papers/ppij.html) Cray Research (http://www.cray.com/) IBM High-Performance Computing (http://ibm.tc.cornell.edu/) ParaSoft Corporation (http://www.parasoft.com/)

2.对于一个100Mb时钟频率的总线连接的SMP系统,假定总线宽度为8个字节,为传输 16个字节的数据,那么总线带宽可达多少, 3.试解释系统带宽与系统互连对剖宽度之间的关系,并举一例说明之。 4.试比较下列5种并行结构的不同点: ①PⅤP:②SMP;③MPP;④COW;⑤DSM 5.试举一例说明: 如何在SIMD机器上实现MMD并行应用? 如何在SIMD模型上实现MPMD并行应用?

2. 对于一个 100Mb 时钟频率的总线连接的 SMP 系统，假定总线宽度为 8 个字节，为传输 16 个字节的数据，那么总线带宽可达多少， 3. 试解释系统带宽与系统互连对剖宽度之间的关系，并举一例说明之。 4. 试比较下列 5 种并行结构的不同点： ①PVP；②SMP；③MPP；④COW；⑤DSM 5. 试举一例说明： 如何在 SIMD 机器上实现 MIMD 并行应用? 如何在 SIMD 模型上实现 MPMD 并行应用?