高性能Python（影印版）

图书标准信息

• 作者 Micha、Lan Ozsvald 著
• 出版社 东南大学出版社
• 出版时间 2015-02
• 版次 1
• ISBN 9787564153854
• 定价 78.00元
• 装帧 平装
• 开本 16开
• 纸张 胶版纸
• 页数 351页
• 字数 455千字
• 正文语种 简体中文
• 原版书名 High Performance Python

【内容简介】

　　你的Python代码也许运行正确，但是你需要运行得更快速。通过探讨隐藏在设计备选方案中的基础理论，《高性能Python（影印版）》将帮助你更深入地理解Python的实现。你将了解如何定位性能瓶颈，从而显著提升高数据流量程序中的代码执行效率。
　　你该如何利用多核架构和集群？或者你该如何搭建一个可以自由伸缩而不会影响可靠性的系统？有经验的Python程序员将会学习到这类问题的具体解决方案，以及来自于各个公司的如何把高性能Python用于社交媒体分析、产品机器学习和其他场景中去的曲折故事。

【作者简介】

作者：（美国）戈雷利克（Micha Gorelick） （英国）欧日沃尔德（Ian Ozsvald） …

【目录】

Preface
1.UnderstandingPerformantPython
TheFundamentalComputerSystem
ComputingUnits
MemoryUnits
CommunicationsLayers
PuttingtheFundamentalElementsTogether
IdealizedComputingVersusthePythonVirtualMachine
SoWhyUsePython？
2.ProfilingtoFindBottlenecks
ProfilingEfficiently
IntroducingtheJuliaSet
CalculatingtheFullJuliaSet
SimpleApproachestoTiming——printandaDecorator
SimpleTimingUsingtheUnixtimeCommand

3.ListsandTuples
AMoreEfficientSearch
ListsVersusTuples
ListsasDynamicArrays
TuplesAsStaticArrays
Wrap-Up

4.DictionariesandSets
HowDoDictionariesandSetsWork？
InsertingandRetrieving
Deletion
Resizing
HashFunctionsandEntropy
DictionariesandNamespaces
Wrap-Up

5.IteratorsandGenerators
IteratorsforInfiniteSeries
LazyGeneratorEvaluation
Wrap-Up

6.MatrixandVectorComputation
IntroductiontotheProblem
Aren'tPythonListsGoodEnough？
ProblemswithAllocatingTooMuch
MemoryFragmentation
Understandingperf
MakingDecisionswithperf'sOutput
Enternumpy
ApplyingnumpytotheDiffusionProblem
MemoryAllocationsandIn-PlaceOperations
SelectiveOptimizations：FindingWhatNeedstoBeFixed
numexpr：MakingIn-PlaceOperationsFasterandEasier
ACautionaryTale：Verify“Optimizations”（scipy）
Wrap-Up

7.CompilingtoC
WhatSortofSpeedGainsArePossible？
JITVersusAOTCompilers
WhyDoesTypeInformationHelptheCodeRunFaster？
UsingaCCompiler
ReviewingtheJuliaSetExample
Cvthon
CompilingaPure-PythonVersionUsingCython
CythonAnnotationstoAnalyzeaBlockofCode
AddingSomeTypeAnnotations
ShedSkin
BuildinganExtensionModule
TheCostoftheMemoryCopies
Cythonandnumpy
ParaUelizingtheSolutionwithOpenMPonOneMachine
Numba
Pythran
PyPy
GarbageCollectionDifferences
RunningPyPyandInstallingModules
WhentoUseEachTechnology
OtherUpcomingProjects
ANoteonGraphicsProcessingUnits（GPUs）
AWishforaFutureCompilerProject
ForeignFunctionInterfaces
ctypes
cffi
f2py
CPythonModule
Wrap-Up

8.Concurrency
IntroductiontoAsynchronousProgramming
SerialCrawler
gevent
tornado
AsyncIO
DatabaseExample
Wrap-Up

9.lhemultiprocessingModule
AnOverviewoftheMultiprocessingModule
EstimatingPiUsingtheMonteCarloMethod
EstimatingPiUsingProcessesandThreads
UsingPythonObjects
RandomNumbersinParallelSystems
Usingnumpy
FindingPrimeNumbers
QueuesofWork
VerifyingPrimesUsingInterprocessCommunication
SerialSolution
NaivePoolSolution
ALessNaivePoolSolution
UsingManager.ValueasaFlag
UsingRedisasaFlag
UsingRawValueasaFlag
UsingmmapasaFlag
UsingmmapasaFlagRedux
SharingnumpyDatawithmultiprocessing
SynchronizingFileandVariableAccess
FileLocking
LockingaValue
Wrap-Up

10.ClustersandJobQueues
BenefitsofClustering
DrawbacksofClustering
$462MillionWallStreetLossThroughPoorClusterUpgradeStrategy
Skype's24-HourGlobalOutage
CommonClusterDesigns
HowtoStartaClusteredSolution
WaystoAvoidPainWhenUsingClusters
ThreeClusteringSolutions
UsingtheParallelPythonModuleforSimpleLocalClusters
UsingIPythonParalleltoSupportResearch
NSQforRobustProductionClustering
Queues
Pub/sub
DistributedPrimeCalculation
OtherClusteringToolstoLookAt
Wrap-Up

11.UsingLessRAM
ObjectsforPrimitivesAreExpensive
TheArrayModuleStoresManyPrimitiveObjectsCheaply
UnderstandingtheRAMUsedinaCollection
BytesVersusUnicode
EfficientlyStoringLotsofTextinRAM
TryingTheseApproacheson8MillionTokens
TipsforUsingLessRAM
ProbabilisticDataStructures
VeryApproximateCountingwitha1-byteMorrisCounter
K-MinimumValues
BloomFilters
LogLogCounter
Real-WorldExample

12.LessonsfromtheField
AdaptiveLab'sSocialMediaAnalytics（SOMA）
PythonatAdaptiveLab
SoMA'sDesign
OurDevelopmentMethodology
MaintainingSoMA
AdviceforFellowEngineers
MakingDeepLearningFlywithRadimRehurek.com
TheSweetSpot
LessonsinOptimizing
Wrap-Up
Large-ScaleProductionizedMachineLearningatLyst.com
PythonsPlaceatLyst
ClusterDesign
CodeEvolutioninaFast-MovingStart-Up
BuildingtheRecommendationEngine
ReportingandMonitoring
SomeAdvice
Large-ScaleSocialMediaAnalysisatSmesh
PythonsRoleatSmesh
ThePlatform
HighPerformanceReal-TimeStringMatching
Reporting，Monitoring，Debugging，andDeployment
PyPyforSuccessfulWebandDataProcessingSystems
Prerequisites
TheDatabase
TheWebApplication
OCRandTranslation
TaskDistributionandWorkers
Conclusion
TaskQueuesatLanyrd.com
Python'sRoleatLanyrd
MakingtheTaskQueuePerformant
Reporting，Monitoring，Debugging，andDeployment
AdvicetoaFellowDeveloper
Index