DirectX, RDX, RSX, and MMX Technology: A Jumpstart Guide to High Performance APIs

Post-publication chapters available! Chapter 24 and Chapter 25 (in PDF format) Until now multimedia developers had to program directly to hardware in order to maximize application performance. DirectX, RDX, RSX, and MMX technology are new advancements that enable programmers to write applications that take advantage of hardware acceleration without direct hardware programming. Written by Intel experts who are developing and applying these new technologies, DirectXAE, RDX, RSX, and MMXoTechnologyo: A Jumpstart Guide to High Performance APIs takes a hands-on approach to illustrate the latest technologies from Microsoft, Intel, and Progressive Networks. This book: *Shows programmers how to get up to speed on each API and provides key hints, tips, and advice throughout the text *Covers DirectX (DirectDraw*, Direct3D*, DirectSound*) and DirectShow (formerly ActiveMovie) APIs from Microsoft; RDX and RSX from Intel; and *RealMedia from Progressive Networks *Illustrates optimization techniques for Pentium, Pentium with MMX *Technology, and the Pentium II processors *Demonstrates how to use Intel's VTune and PMonitor for processor and memory optimization 0201309440B04062001

Table of Contents:
Preface. Introduction. I. SURVEYING MULTIMEDIA. 1. Overview of Media on the PC. Background. Graphics Device Independence. Motion Video under Windows. Multimedia Gaming under Windows 95. 3D Video Architectures on the PC. Audio Architectures on the PC. 2. Processor Architecture Overview. Processor Architecture. System Overview. II. SPRITES, BACKGROUNDS, AND PRIMARY SURFACES. 3. Simple Sprites in GDI. Graphics Device Interface (GDI) Overview. Animation Objects. Sprites. Backgrounds. Transparent Blts with GDI. Backgrounds. Demo Time. How Fast Does GDI Draw Sprites and Backgrounds? 4. Sprites with a DirectDraw Primary Surfaces. Introduction to Microsoft’s DirectDraw. Features of DirectDraw. Before You Get Overly Excited. Instantiating a DirectDraw Object. Querying and Creating a Primary Surface. Implementing a Simple Sprite Class. Drawing a Sprite on the DirectDraw Primary Surface. Demo Time. Redrawing Backgrounds on a DirectDraw Primary Surface. How Fast Can We Draw Sprites and Backgrounds? Compositing Objects on a DirectDraw Primary Surface. 5. Hardware Acceleration via DirectDraw. Creating an Offscreen Surface. Drawing a Sprite on the DirectDraw Offscreen Surface. Demo Time. How Fast is OffScreen Surface Drawing? Finding Hardware Acceleration. Setting Up for Hardware Acceleration. How Fast is CSurfaceVidMem Drawing? Accelerating Offscreen to Primary Transfers by Page Flips. What is Graphics Page Flipping? DirectDraw Page Flipping Model. Does the Hardware Support Page Flipping? Setting Up DirectDraw to Use Page Flipping. “Rendering” Flippable Surfaces. How Fast is CSurfaceBackBuffer Drawing? Hardware Acceleration to Blt Sprites. How Fast is CSpriteGrfx (and CBackgroundGrfx) Drawing? 6. RDX: High-Performance Mixing with a High-Level API. Introduction to Intel’s RDX Animation Library. Features of RDX. Before You Get Overly Excited. Using RDX. Generic Objects with RDX. The Programming Model. Working with RDX. Creating an RDX Surface. An RDX Sprite Class. Drawing the RDX Sprite. Demo Time. How Fast Does CSurfaceRdx Draw? Hardware Acceleration with RDX. Full Screen Mode with RDX. How Fast Does CSurfaceRdx Draw in Full Screen Mode? Accelerating Objects with RDX. III. MAKING THE MEDIA MIX. 7. Video under Windows. Concepts of Motion Video. Capturing and Compressing Video. Windows Multimedia Architectures. Overview of Video Codecs. 8. DirectShow Filters. DirectShow Components. What’s a Filter Graph? Understanding the Mighty Filter. An Overview on the Samples. Creating a Source Filter. The Source Filter Class. Create an Instance of the Source Filter. The Source Stream Class. The Connection Process. Starting and Stopping. Moving the Data. Creating a Transform Filter. Creating a Rendering Filter. Adding Your Own Interface. Adding Property Pages to Filters. Adding the Property Interface to the Filter. Implementing the Property Page Interface. Adding a Filter to the Registry. Using a Registry File is Not Recommended. Using Filter Self-Registration is Recommended. 9. DirectShow Applications. DirectShow Mechanisms for Working on Filter Graphs. COM: Automatic Construction of Filter Graphs. COM: Manual Construction of Filter Graphs. Adding Source Filters. Adding Transform and Rendering Filters. Connecting the Two Pins. COM: Accessing Custom Interfaces. COM: Showing Filter Property Pages. Creating Events under DirectShow. ActiveX: A Simple Way to Control DirectShow. Playing a File Using the ActiveX Interface. Controlling the ActiveX Control from Your Application. ActiveX: Handling Events. 10. Mixing Sprites, Backgrounds, and Videos. Introduction to Mixing. Mixing Sprites with Video. Mixing Animation with Video. Mixing with RDX. Playing Video with the RDX DirectShow Interface. Mixing a Sprite on Top of Video. Mixing Video on Video. 11. Streaming down the Superhighway with RealMedia. Overview of RealMedia. The RealMedia Plug-in Architecture. Data Flows: Server to Client. Data Management Objects. IRMABuffer: Dynamic Memory Allocation Object. IRMAValues: Indexed List Object. IRMAPacket: Packet Transport Object. RealMedia Asynchronous Interfaces. Common Requirements for All Plug-ins. Building a File-Format Plug-in. Initializing the File-Format Plug-in. File and Stream Headers. Let the Streaming Begin! Building a Rendering Plug-in. RealMedia Audio Services. Playing a Simple Pulse Coded Modulation (PCM) Audio File. Pump Up the Volume. IV. PLAYING AND MIXING SOUND WITH DIRECTSOUND AND RSX 3D. 12. Audio Mixing with DirectSound. Overview of Audio under Windows 95. DirectSound Features. DirectSound Architecture. Playing a WAV File Using DirectSound. Initializing DirectSound. DirectSound Structures. Creating Sound Buffers. Playing the Sound. Demo Time. Mixing Two WAV Files. Controlling the Primary Sound Buffer. Initializing to Get Control of the Output Format. Creating a Primary DirectSound Buffer. Demo Time. 13. Realistic 3D Sound Experience: RSX 3D. RSX 3D Features. Creating an RSX 3D Object. Creating the Cached Emitter and the Direct Listener Objects. Play One WAV File. Mixing Many WAV Files. RSX Goes 3DNTrue 3D Sound. Setting Up 3D Sound with RSX 3D. Adding Special Sound Effects with RSX 3D. The Doppler Effect. The Reverberation Effect. Audio Streaming in RSX 3D. V. WELCOME TO THE THIRD DIMENSION. 14. An Introduction to Direct3D. Some Background on 3D on the PC. Introduction to Direct3D. A Taste of Direct3D’s Retained Mode. Direct3D’s Immediate Mode. Before You Get Overly Excited. Inside Direct3D. Direct3D and DirectDraw. Direct3D Rendering Engine. Revving Up Direct3D. The Starting Point: IDirect3D Object. Enumerating IDirect3DDevices. Creating an IDirect3DDevice. Preparing a Palette. Extending the Surface for 3D. Mapping from a 3D Model to the 2D Surface Using Viewports. Talking to 3D Devices Through Execute Buffers. Execute Operations. Operations Used to Render a Simple Triangle. Executing the Execute Buffers. Seeing Results from 3D Devices. Demo Time. 15. Embellishing Our Triangle with Backgrounds, Shading, and Textures. Continuing Our Look into Direct3D. Repainting the Background Using Direct3D. Looking at Direct3D Materials. Creating a Direct3D Background. Bltting a Direct3D Background. Controlling Shading Options. Looking at Some Render States and Their Default Values. Coloring a Pixel in Direct3D. Shading with the RGB Color Model. Shading with the Ramp Color Model. Changing Default Render States. Texture Mapping with Direct3D. Creating a Texture Map. Setting Up Triangle Vertices for Texture Mapping. Setting Up Render Operations for Texture Mapping. Handling “Lit” Texture Maps. Z-Buffering with Direct3D. Why Bother with Z-Buffering? Setting Up for Z-Buffering. 16. Understanding and Enhancing Direct3D Performance. How Fast Does Our Triangle Run? Stages of Rendering Our Triangle. Measuring the Rendering Stages of Our Triangle. Trimming Some Fat from the Rendering Stages. Measuring Shading Options. Measuring the Performance of Shading Options in Our Triangle. Measuring the Performance of Texture-Mapping in Our Triangle. Adding a Z-Buffer to the Recipe. Getting Perspective: Comparing 3D (RGB Mode) to 2D. Improving Performance Using the Ramp Driver. Loading the Ramp Color Model Driver. Using the Ramp DriverNThe First Try. Creating Materials for the Ramp Driver. Rendering a Triangle with the Ramp Driver. How Does the Ramp Driver Perform? Optimizing Texture Mapping. 17. Mixing 3D with Sprites, Backgrounds, and Videos. Mixing a 3D Object on a 2D Background. Our 3D Surface is Also a 2D Surface. Measuring Background Performance. Mixing in Sprites. Using RDX to Mix in Sprites. Adding RDX Objects at Front and Back. Mixing in Video. Handling Palettes. Using Video as a Texture Map. VI. PROCESSORS AND PERFORMANCE OPTIMIZATION. 18. The Pentium Processor Family. Basic Concepts and Terms. The Pentium Processors. The Pentium Processor. The Pentium Pro Processor. The Pentium Processor with MMX Technology. The Pentium II Processor. Identifying Processor Models. 19. The Pentium Processor. Architectural Overview. Instruction and Data L1 Caches. Operational Overview. Performance Considerations. Instruction Prefetch. Operational Overview. Performance Considerations. Branch Prediction and the Branch Target Buffer. Operational Overview. A Closer Look at the BTB. Performance Considerations. Dual Pipelined Execution. Operational Overview. Performance Considerations. Pentium Integer Pairing Rules. Address Generation Interlock (AGI). Write Buffers. Operational Overview. Performance Considerations. Revisiting Our Sprite Sample. Overview of the Assembly Version of CSprite. Analyzing the Performance of Our Sprite Sample. Do I Really Need to Schedule My Code? 20. The Pentium Processor with MMX Technology. A Look at MMX Technology. SIMD. Architectural Overview. The Pool of Four Write Buffers. MMX Uses Floating-Point Registers. EMMS to the Rescue: How to Mix MMX and FP Instructions. MMX Technology Data Types. The MMX Instruction Set. Using MMX Technology to Render Our Sprite Sample. MMX Technology Optimization Rules and Penalties. MMX Exceptions to General Pentium Rules. MMX Instruction Pairing Rules. MMX Instruction Scheduling Rules. Performance Analysis of Our Sprite. MMX versus Integer Implementation of the Sprite. 21. VTune and Other Performance Optimization Tools. Overview of Performance Counters. Introducing VTune. VTune Static Analysis. Tune Dynamic Analysis. Systemwide MonitoringNTime-and Event-Based Sampling. Read Time Stamp Counter. The PMonitor Event Counter Library. 22. The Pentium II Processor. Architectural Overview. The Life Cycle of an Instruction on the Pentium II. Comparing the Pentium II with the Pentium Pro Processor. Comparing the Pentium II with the Pentium with MMX Technology Processor. Instruction and Data Caches. Operational Overview. Performance Considerations. Instruction Fetch Unit. Operational Overview. Performance Considerations. Fetch Performance with Event Counters. Branch Prediction and the Branch Target Buffer. Operational Overview. Performance Considerations. Branch Performance with Event Counters. Instruction Decoders. Operational Overview. Performance Considerations. Register Alias Table Unit. Operational Overview. Performance Considerations. Reorder Buffer and Execution Units. Operational Overview. Performance Considerations. Retirement Unit. Rendering Our Sprite on the Pentium II. Speed Up Graphics Writes with Write Combining. Operational Overview. Performance Considerations. 23. Memory Optimization: Know Your Data. Overview of the Memory Subsystem. Architectural Overview. Memory Pages and Memory Access Patterns. Memory Timing. Performance Considerations. Architectural Differences among the Pentium and Pentium Pro Processors. Architectural Cache Differences. Write Buffer Differences. Data Controlled Unit Splits on the Pentium Pro Processor. Partial Memory Stalls. Maximizing Aligned Data and MMX Stack Accesses. The Pitfalls of Unaligned MMX Stack Access. Accessing Cached Memory. Writing to Video Memory. Using Aligned Accesses to Video Memory. Spacing Out Writes to Video Memory with Write Buffers. The Finale. The Spiral Continues. The Hardware Spiral. The Software Spiral. Remote Multimedia (a.k.a. Internet Multimedia). Internet Languages. Multimedia on the Internet. Evolving Hardware for the Internet. Multimedia Conferencing. Better, Faster, Cheaper 3D. 3D Hardware Spiral. 3D Software Spiral. 3D Scalability. Emerging Application Areas. Multimedia in the Home. Demo Time. Index. 0201309440T04062001

About the Author :
Rohan Coelho holds degrees in physics and computer science. For most of his eight years at Intel he has specialized in multimedia technologies at the Intel Architecture Labs (IAL). He wrote the first Indeo Video decoder, participated with Microsoft in developing VFW and ActiveMovie, co-architected DCI for Windows 3.1, worked with Sega on their SonicPC product, architected RDX, and optimized 3D rendering for MMX technology. He has published several papers and holds multiple patents. Maher Hawash has been a multimedia software developer at Intel for the past five years. He graduated with an MSEE from the University of Texas at Arlington. As a lead engineer on the MMX technology software team, he developed the MMX technology emulator and optimized MPEG decoders. As part of the Intel Architecture Labs (IAL), Maher focuses on video technologies, including VFW, ActiveMovie, Indeo, ProShare video conferencing, and the Intel Smart Video Recorder. 0201309440AB04062001