并行程序设计导论：（英文版）

图书标准信息

• 作者 [美]帕切克 著
• 出版社 机械工业出版社
• 出版时间 2011-11
• 版次 1
• ISBN 9787111358282
• 定价 65.00元
• 装帧 平装
• 开本 16开
• 纸张 胶版纸
• 页数 370页
• 正文语种 英语
• 原版书名 An Introduction to Parallel Programming

【内容简介】

采用教程形式，从简短的编程实例起步，一步步编写更有挑战性的程序。重点介绍分布式内存和共享式内存的程序设计、调试和性能评估。使用MPI、PTrlread和OperIMP等编程模型，强调实际动手开发并行程序。并行编程已不仅仅是面向专业技术人员的一门学科。如果想要全面开发机群和多核处理器的计算能力，那么学习分布式内存和共享式内存的并行编程技术是不可或缺的。由PeterS.Pacheco编著的《并行程序设计导论（英文版）》循序渐进地展示了如何利用MPI、PThread和OperlMP开发高效的并行程序，教给读者如何开发、调试分布式内存和共享式内存的程序，以及对程序进行性能评估。

【作者简介】

帕切克（PetmS.Pacheco），拥有佛罗里达州立大学数学专业博士学位。曾担任旧金山大学计算机主任，目前是旧金山大学数学系主任。近20年来，一直为本科生和研究生讲授并行计算课程。

【目录】

CHAPTER1WhyParallelComputing?
1.1WhyWeNeedEver-IncreasingPerformance
1.2WhyWe'reBuildingParallelSystems
1.3WhyWeNeedtoWriteParallelPrograms
1.4HowDoWeWriteParallelPrograms?
1.5WhatWe'llBeDoing
1.6Concurrent，Parallel，Distributed
1.7TheRestoftheBook
1.8AWordofWarning
1.9TypographicalConventions
1.10Summary
1.11Exercises

CHAPTER2ParallelHardwareandParallelSoftware
2.1SomeBackground
2.1.1ThevonNeumannarchitecture
2.1.2Processes，multitasking，andthreads
2.2ModificationstothevonNeumannModel
2.2.1Thebasicsofcaching
2.2.2Cachemappings
2.2.3Cachesandprograms:anexample
2.2.4Virtualmemory
2.2.5Instruction-levelparallelism
2.2.6Hardwaremultithreading.
2.3ParallelHardware
2.3.1SIMDsystems
2.3.2MIMDsystems
2.3.3Interconnectionnetworks
2.3.4Cachecoherence
2.3.5Shared-memoryversusdistributed-memory
2.4ParallelSoftware
2.4.1Caveats
2.4.2Coordinatingtheprocesses/threads
2.4.3Shared-memory
2.4.4Distributed-memory
2.4.5Programminghybridsystems
2.5InputandOutput
2.6Performance
2.6.1Speedupandefficiency
2.6.2Amdahl'slaw
2.6.3Scalability
2.6.4Takingtimings
2.7ParallelProgramDesign
2.7.1Anexample
2.8WritingandRunningParallelPrograms
2.9Assumptions
2.10Summary
2.10.1Serialsystems
2.10.2Parallelhardware
2.10.3Parallelsoftware
2.10.4Inputandoutput
2.10.5Performance.
2.10.6Parallelprogramdesign
2.10.7Assumptions
2.11Exercises

CHAPTER3Distributed-MemoryProgrammingwithMPI
3.1GettingStarted
3.1.1Compilationandexecution
3.1.2MPIprograms
3.1.3MPIInitandMPIFinalize
3.1.4Communicators，MPICommsizeandMPICommrank
3.1.5SPMDprograms
3.1.6Communication
3.1.7MPISend
3.1.8MPIRecv
3.1.9Messagematching
3.1.10Thestatuspargument
3.1.11SemanticsofMPISendandMPIRecv
3.1.12Somepotentialpitfalls
3.2TheTrapezoidalRuleinMPI
3.2.1Thetrapezoidalrule
3.2.2Parallelizingthetrapezoidalrule
Contentsxiii
3.3DealingwithI/O
3.3.1Output
3.3.2Input
3.4CollectiveCommunication
3.4.1Tree-structuredcommunication
3.4.2MPIReduce
3.4.3Collectivevspoint-to-pointcommunications
3.4.4MPIAllreduce
3.4.5Broadcast
3.4.6Datadistributions
3.4.7Scatter
3.4.8Gather
3.4.9Allgather
3.5MPIDerivedDatatypes
3.6PerformanceEvaluationofMPIPrograms
3.6.1Takingtimings
3.6.2Results
3.6.3Speedupandefficiency
3.6.4Scalability
3.7AParallelSortingAlgorithm
3.7.1Somesimpleserialsortingalgorithms
3.7.2Parallelodd-eventranspositionsort
3.7.3SafetyinMPIprograms
3.7.4Finaldetailsofparallelodd-evensort
3.8Summary
3.9Exercises
3.10ProgrammingAssignments.

CHAPTER4Shared-MemoryProgrammingwithPthreads.
4.1Processes，Threads，andPthreads
4.2Hello，World
4.2.1Execution
4.2.2Preliminaries
4.2.3Startingthethreads
4.2.4Runningthethreads
4.2.5Stoppingthethreads
4.2.6Errorchecking
4.2.7Otherapproachestothreadstartup
4.3Matrix-VectorMultiplication
4.4CriticalSections
xivContents
4.5Busy-Waiting
4.6Mutexes.
4.7Producer-ConsumerSynchronizationandSemaphores
4.8BarriersandConditionVariables
4.8.1Busy-waitingandamutex
4.8.2Semaphores
4.8.3Conditionvariables
4.8.4Pthreadsbarriers
4.9Read-WriteLocks
4.9.1Linkedlistfunctions
4.9.2Amulti-threadedlinkedlist
4.9.3Pthreadsread-writelocks
4.9.4Performanceofthevariousimplementations
4.9.5Implementingread-writelocks
4.10Caches，CacheCoherence，andFalseSharing
4.11Thread-Safety
4.11.1Incorrectprogramscanproducecorrectoutput
4.12Summary
4.13Exercises
4.14ProgrammingAssignments.

CHAPTER5Shared-MemoryProgrammingwithOpenMP.
5.1GettingStarted
5.1.1CompilingandrunningOpenMPprograms
5.1.2Theprogram
5.1.3Errorchecking
5.2TheTrapezoidalRule
5.2.1AfirstOpenMPversion
5.3ScopeofVariables
5.4TheReductionClause.
5.5TheparallelforDirective
5.5.1Caveats
5.5.2Datadependences
5.5.3Findingloop-carrieddependences
5.5.4Estimating
5.5.5Moreonscope
5.6MoreAboutLoopsinOpenMP:Sorting.
5.6.1Bubblesort
5.6.2Odd-eventranspositionsort
5.7SchedulingLoops
5.7.1Thescheduleclause
5.7.3Thedynamicandguidedscheduletypes
5.7.4Theruntimescheduletype
5.7.5Whichschedule?
5.8ProducersandConsumers
5.8.1Queues
5.8.2Message-passing
5.8.3Sendingmessages
5.8.4Receivingmessages
5.8.5Terminationdetection
5.8.6Startup
5.8.7Theatomicdirective
5.8.8Criticalsectionsandlocks
5.8.9Usinglocksinthemessage-passingprogram
5.8.10criticaldirectives，atomicdirectives，orlocks?
5.8.11Somecaveats
5.9Caches，CacheCoherence，andFalseSharing
5.10Thread-Safety
5.10.1Incorrectprogramscanproducecorrectoutput
5.11Summary
5.12Exercises
5.13ProgrammingAssignments.

CHAPTER6ParallelProgramDevelopment
6.1Twon-BodySolvers
6.1.1Theproblem
6.1.2Twoserialprograms
6.1.3Parallelizingthen-bodysolvers
6.1.4AwordaboutI/O
6.1.5ParallelizingthebasicsolverusingOpenMP
6.1.6ParallelizingthereducedsolverusingOpenMP
6.1.7EvaluatingtheOpenMPcodes
6.1.8Parallelizingthesolversusingpthreads
6.1.9ParallelizingthebasicsolverusingMPI
6.1.10ParallelizingthereducedsolverusingMPI
6.1.11PerformanceoftheMPIsolvers
6.2TreeSearch
6.2.1Recursivedepth-firstsearch
6.2.2Nonrecursivedepth-firstsearch
6.2.3Datastructuresfortheserialimplementations
6.2.6Astaticparallelizationoftreesearchusingpthreads
6.2.7Adynamicparallelizationoftreesearchusingpthreads
6.2.8Evaluatingthepthreadstree-searchprograms
6.2.9Parallelizingthetree-searchprogramsusingOpenMP
6.2.10PerformanceoftheOpenMPimplementations
6.2.11ImplementationoftreesearchusingMPIandstatic
partitioning
6.2.12ImplementationoftreesearchusingMPIanddynamic
partitioning
6.3AWordofCaution
6.4WhichAPI?
6.5Summary
6.5.1PthreadsandOpenMP
6.5.2MPI
6.6Exercises
6.7ProgrammingAssignments
CHAPTER7WheretoGofromHere
References
Index