MATLAB性能提升1001计：速度优化秘籍

需积分: 10 168 浏览量更新于2024-07-16 收藏 37.82MB PDF 举报

"《加速MATLAB性能：1001个提升代码运行速度的技巧》是一本由Yair Altman编写的实用指南，专门针对MATLAB和Simulink用户。本书旨在帮助读者解决在使用MATLAB编程过程中遇到的性能瓶颈问题。作者提供了1001条详实的建议和实例，涵盖了多个关键领域，如代码分析、并行计算、GPU加速、隐式矢量化并行操作以及有效的内存管理。首先，本书强调了代码性能优化的重要性，尤其是在处理大规模数据或复杂的数学运算时，提高运行速度可以显著提升工作效率。作者深入探讨了如何通过分析代码结构，识别出潜在的效率低下部分，例如循环迭代和函数调用的优化，以及避免不必要的计算。其次，书中介绍了如何利用MATLAB内置的并行计算工具箱，如parfor循环和并行池，来实现多核处理器的并行执行，从而大大减少计算时间。同时，对于GPU加速，MATLAB提供了GPU Coder工具，让开发者能够将部分任务移至图形处理器上运行，以利用其强大的并行处理能力。此外，隐式矢量化是MATLAB的一大特性，通过利用向量化操作，可以避免显式地创建临时数组，从而减少内存消耗。作者展示了如何有效地利用这一特性，提高代码的执行效率。内存管理方面，书中提到了诸如clear、delete和rehashing等方法，以减少内存占用和提升程序的响应速度。书中还包含了与MATLAB版本更新相关的最佳实践，因为新版本往往引入了性能改进和新的功能，充分利用这些特性也能提升代码的运行速度。此外，书中还会讨论如何在实际项目中平衡性能优化与代码可读性和维护性，确保在追求速度的同时，不会牺牲代码的清晰度和复用性。《加速MATLAB性能：1001个提升代码运行速度的技巧》不仅提供了一种工具集，更是一种思考方式，引导读者从多个角度去审视和优化自己的MATLAB代码，使其在各种场景下都能保持高效运行。这本书对于所有希望通过提升MATLAB性能来优化工作流程的专业人士来说，都是一份不可或缺的参考资料。"

xiv Contents

7.2 Alternative/Related Technologies .......................................................................... 364

7.2.1 Using GPUmat ..............................................................................................36 4

7.2.2 Multicore Library for Parallel Processing on Multiple Cores ................368

7.2.3 Using ArrayFire Library via MEX Interface ............................................ 370

7.2.4 Additional Alternatives .............................................................................. 373

7.3 Multithreading .......................................................................................................... 374

7.3.1 Using POSIX Threads .................................................................................. 374

7.3.2 Using OpenMP ............................................................................................. 377

7.3.3 Using Java Threads ...................................................................................... 379

7.3.4 Using .Net Threads ...................................................................................... 382

7.3.5 Using MATLAB Timers ..............................................................................384

7.4 Spawning External Processes..................................................................................385

8 Using Compiled Code ........................................................................................................389

8.1 Using MEX Code .......................................................................................................389

8.1.1 Introduction .................................................................................................. 389

8.1.2 Our First MEX Function .............................................................................390

8.1.3 MEX Function Inputs and Outputs ........................................................... 39 4

8.1.4 Accessing MATLAB Data ........................................................................... 395

8.1.5 A Usage Example ......................................................................................... 401

8.1.6 Memory Management ................................................................................. 404

8.1.7 Additional Aspects ...................................................................................... 406

8.2 Using the MATLAB Coder Toolbox ....................................................................... 413

8.2.1 Code Adaptation .......................................................................................... 4 14

8.2.2 A Simple Example: Euclidean Distances Algorithm .............................. 416

8.2.3 A More Realistic Example: Dijkstra’s Shortest-Path Algorithm ............ 420

8.2.4 Conguring the Coder for Maximal Performance .................................427

8.3 Porting MATLAB Algorithms to FPGA.................................................................433

8.3.1 Algorithm Adaptation .................................................................................434

8.3.2 HDL Workow ............................................................................................. 435

8.3.3 Run-Time Measurements ............................................................................444

8.4 Deployed (Compiled) MATLAB Programs ...........................................................445

8.5 Using External Libraries ..........................................................................................450

8.5.1 Introduction .................................................................................................. 450

8.5.2 Java ................................................................................................................. 451

8.5.3 NAG: Numerical Algorithms Group.........................................................452

8.5.4 MATLAB Toolbox ........................................................................................455

8.5.5 MCT: Multi-Precision Computing Toolbox ..............................................456

8.5.6 Additional Libraries ....................................................................................457

9 Memory-Related Techniques ...........................................................................................459

9.1 Why Memory Affects Performance ........................................................................ 459

9.2 Proling Memory Usage .......................................................................................... 461

9.2.1 Workspace Browser ..................................................................................... 461

9.2.2 whos Function .............................................................................................. 4 62

9.2.3 memory Function ........................................................................................463

9.2.4 feature memstats and feature dumpmem ...............................................46 4

9.2.5 feature mtic/mtoc ........................................................................................466

9.2.6 Proler’s Memory-Monitoring Feature ..................................................... 467

xvContents

9.2.7 Proling Java Memory ................................................................................ 469

9.2.8 Using Third-Party Tools ............................................................................. 470

9.2.9 Using the Operating System’s Tools ..........................................................47 2

9.2.10 format debug ................................................................................................ 473

9.3 MATLAB’s Memory Storage and Looping Order ................................................ 475

9.3.1 Memory Storage of MATLAB Array Data ................................................ 475

9.3.2 Loop Down Columns Rather than Rows .................................................477

9.3.3 Effect of Subindexing ..................................................................................480

9.4 Array Memory Allocation ....................................................................................... 481

9.4.1 Dynamic Array Growth .............................................................................. 481

9.4.2 Effects of Incremental JIT Improvements ................................................. 482

9.4.3 Preallocate Large Data Arrays ...................................................................483

9.4.4 Preallocation Location within the Code ................................................... 488

9.4.5 Preallocating Nondouble Data ...................................................................488

9.4.6 Alternatives for Enlarging Arrays ............................................................. 491

9.5 Minimizing Memory Allocations ........................................................................... 493

9.5.1 MATLAB’s Copy-on-Write Mechanism .................................................... 493

9.5.2 In-Place Data Manipulations ...................................................................... 496

9.5.3 Reusing Variables (with Utmost Care) ......................................................5 02

9.5.4 Clearing Unused Workspace Variables ....................................................505

9.5.5 Global and Persistent Variables .................................................................507

9.5.6 Scoping Rules and Nested Functions .......................................................509

9.5.7 Passing Handle References (Not Data) to Functions .............................. 5 11

9.5.8 Reducing Data Precision/Type .................................................................. 512

9.5.9 Devectorizing Huge Data Operations ...................................................... 513

9.5.10 Assign Anonymous Functions in Dedicated Wrapper Functions ........ 514

9.5.11 Represent Objects by Simpler Data Types ................................................ 516

9.6 Memory Packing ....................................................................................................... 518

9.7 Additional Recommendations ................................................................................ 519

9.7.1 MATLAB Variables ...................................................................................... 5 19

9.7.2 Java Objects ................................................................................................... 52 2

10 Graphics and GUI ...............................................................................................................52 5

10.1 Initial Graphs Generation ........................................................................................ 526

10.1.1 Reduce the Number of Plotting Elements ................................................ 526

10.1.2 Use Simple or No Plot Markers ..................................................................530

10.1.3 Use Vectorized Data for Plotting ............................................................... 532

10.1.4 Use Static Axes Properties .......................................................................... 533

10.1.5 Only Use drawnow when You Are Finished Plotting ........................... 534

10.1.6 Use Low-Level Rather than High-Level Graphic Functions .................536

10.1.7 Generate Plots while the Figure Is Hidden .............................................. 537

10.1.8 Apply Data Reduction to Plotted Data to Fit the Display Area ............538

10.1.9 Reuse Plot Axes ............................................................................................ 540

10.1.10 Avoid Using the axes Function ..................................................................540

10.1.11 Use the Painters Figure Renderer with Fast Axes DrawMode ..............542

10.1.12 Images ............................................................................................................543

10.1.13 Patches and Volume Surfaces .....................................................................5 45

10.1.14 Colorbars ....................................................................................................... 546

10.1.15 Legends .........................................................................................................546

xvi Contents

10.1.16 Reopen Presaved Figure and/or Plot Axes ..............................................547

10.1.17 Set Axes SortMethod to Childorder and Reduce Transparency ............5 48

10.2 Updating Graphs and Images in Real Time ..........................................................549

10.2.1 Axes Update .................................................................................................. 549

10.2.2 Plot and Image Update ................................................................................550

10.2.3 Legends and Colorbars ............................................................................... 552

10.2.4 Accessing Object Properties .......................................................................553

10.2.5 Listeners and Callbacks ..............................................................................554

10.2.6 Trading Accuracy for Speed .......................................................................555

10.2.7 Avoid Update to the Same Value ............................................................... 557

10.2.8 Cache Graphic Handles .............................................................................. 557

10.2.9 Avoid Interlacing Property get and set ....................................................558

10.2.10 Use hgtransform to Transform Graphic Objects ....................................559

10.3 Figure Window Performance Aspects ................................................................... 559

10.3.1 Use Hardware-Accelerated OpenGL Renderer and Functionality ...... 560

10.3.2 Set a Nondefault WVisual/XVisual Property Value .............................. 561

10.3.3 Disable BackingStore ...................................................................................5 62

10.3.4 Disable DoubleBuffer .................................................................................. 562

10.3.5 Set a Manual DitherMapMode on Old Platforms ...................................563

10.3.6 Reuse Figure Windows ...............................................................................563

10.3.7 Sharing Data between GUI Callback Functions .....................................564

10.3.8 Disable Anti-Aliasing .................................................................................. 564

10.3.9 Use Smaller and Fewer Figure Windows ................................................. 564

10.4 GUI Preparation and Responsiveness ....................................................................565

10.4.1 Creating the Initial GUI ..............................................................................565

10.4.2 Presenting User Feedback .......................................................................... 575

10.4.3 Performing Asynchronous Actions ..........................................................584

10.5 Avoiding Common Pitfalls ...................................................................................... 591

10.5.1 Minimize Intentional Pauses ..................................................................... 591

10.5.2 Delete Unused Graphic Objects ................................................................. 592

11 I/O Techniques .................................................................................................................... 593

11.1 Reducing the Amount of I/O .................................................................................. 593

11.2 Avoiding Repeated File Access ............................................................................... 595

11.3 Reading and Writing Files ....................................................................................... 597

11.3.1 Text versus Binary Format .......................................................................... 597

11.3.2 Text File Pre-Processing .............................................................................. 598

11.3.3 Memory-Mapped Files ................................................................................ 598

11.3.4 Reading Files Efciently .............................................................................600

11.3.5 Writing Files Efciently ..............................................................................604

11.4 Data Compression and the save Function ............................................................ 607

11.5 Excel Files (and Microsoft Ofce Files in General) .............................................. 614

11.6 Image Files ................................................................................................................. 620

11.7 Using Java and C I/O ................................................................................................622

11.8 Searching, Parsing, and Comparing Files ............................................................. 625

11.8.1 Searching for Files ....................................................................................... 625

11.8.2 Parsing and Scanning Files ........................................................................ 626

11.9 Additional Aspects ................................................................................................... 627

11.9.1 Using p-Code ................................................................................................627

剩余766页未读，继续阅读

huashi20130128

粉丝: 0
资源: 3

MATLAB性能提升1001计：速度优化秘籍

优化MATLAB性能：1001个加速技巧

提升MATLAB性能的全方位技术指南

"大规模游戏社交网络节点相似性算法及其在云上深度学习训练中的应用

Accelerating MATLAB Performance 1001 tips to speed up MATLAB programs

Accelerating MATLAB Performance 1001 Tips to Speed Up MATLAB Program

Accelerating MATLAB Performance - 1001 Tips to Speed Up MATLAB Programs

《Accelerating MATLAB Performance - 1001 Tips to Speed Up MATLAB Programs》

MATLAB 加速方法大全 Accelerating MATLAB Performance

MATLAB Genetic Algorithm Parallel Computing Guide: Accelerating Optimization, Saving Time

MATLAB Matrix Parallel Computing: Leveraging Multi-core Advantages to Boost Computing Speed, A Three...

最新资源