Intel 64位汇编优化手册：加速程序性能

5星 · 超过95%的资源需积分: 9 72 浏览量更新于2024-08-01 收藏 3.57MB PDF 举报

本资源是Intel 64位和IA-32架构的优化参考手册，编号为248966-019，发布于2009年10月。该文档旨在帮助开发者深入了解和优化在Intel 64位架构下的汇编编程，以提升程序运行效率。它涵盖了针对Intel处理器的高级汇编语言优化技巧，包括但不限于指令集、内存管理、数据类型处理、向量化操作和SIMD（单指令多数据）技术。文档强调了Intel不授予任何与产品相关的知识产权许可，除非明确在销售条款中有所规定。用户应清楚，Intel对产品的提供不承担任何责任，包括但不限于特定用途的适用性、产品品质或对专利、版权或其他知识产权的侵犯。此外，Intel提醒开发者，其产品并非设计用于可能导致人身伤害或死亡的场合，因此在使用时需谨慎考虑安全因素。在技术内容方面，手册详细说明了如何避免使用未定义或预留的指令，因为Intel可能随时更新规格和产品描述。这意味着开发者在编写代码时应密切关注官方文档，确保他们的优化策略不会与未来版本的处理器兼容性产生冲突。优化指南可能涉及以下核心知识点： 1. **指令集优化**：介绍不同类型的Intel 64位指令，如SSE（Streaming SIMD Extensions）、AVX（Advanced Vector Extensions）等，以及如何利用这些特性进行高效计算。 2. **内存管理优化**：讲解如何有效地管理64位地址空间，包括缓存优化、内存布局和数据结构设计，以减少内存访问延迟。 3. **数据类型和向量化**：阐述如何利用宽字节数据类型和SIMD指令处理大量数据，提高并行性和性能。 4. **控制流优化**：探讨循环展开、分支预测和无条件转移的使用，以避免控制流的分支开销。 5. **错误处理和异常处理**：介绍如何编写健壮的代码，以减少中断和异常带来的性能损失。 6. **编译器和汇编器特性**：解释如何利用Intel编译器和汇编器提供的优化选项，如编译器的指令集选择和循环矢量化功能。 7. **平台依赖性**：警告开发者注意特定处理器特有的特性和限制，确保跨平台代码的兼容性。这份文档对于任何希望在Intel 64位系统上实现高性能汇编编程的开发者来说，是一份不可或缺的参考资料，它提供了深入的理论指导和技术实践案例，能够显著提升程序的运行速度和效率。

CONTENTS

xvi

PAGE

EXAMPLES

Example 3-1. Assembly Code with an Unpredictable Branch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-8

Example 3-2. Code Optimization to Eliminate Branches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-8

Example 3-4. Use of PAUSE Instruction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9

Example 3-3. Eliminating Branch with CMOV Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9

Example 3-5. Pentium 4 Processor Static Branch Prediction Algorithm . . . . . . . . . . . . . . . . . . . . . . 3-10

Example 3-6. Static Taken Prediction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11

Example 3-7. Static Not-Taken Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11

Example 3-8. Indirect Branch With Two Favored Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-14

Example 3-9. A Peeling Technique to Reduce Indirect Branch Misprediction . . . . . . . . . . . . . . . . . 3-15

Example 3-10. Loop Unrolling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-16

Example 3-11. Macro-fusion, Unsigned Iteration Count . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-19

Example 3-12. Macro-fusion, If Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-20

Example 3-13. Macro-fusion, Signed Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-21

Example 3-14. Macro-fusion, Signed Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-21

Example 3-15. Avoiding False LCP Delays with 0xF7 Group Instructions . . . . . . . . . . . . . . . . . . . . . . 3-23

Example 3-16. Clearing Register to Break Dependency While Negating Array Elements. . . . . . . . 3-29

Example 3-17. Spill Scheduling Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-32

Example 3-18. Dependencies Caused by Referencing Partial Registers . . . . . . . . . . . . . . . . . . . . . . . 3-35

Example 3-19. Avoiding Partial Register Stalls in Integer Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-35

Example 3-20. Avoiding Partial Register Stalls in SIMD Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-36

Example 3-21. Avoiding Partial Flag Register Stalls. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-37

Example 3-22. Reference Code Template for Partially Vectorizable Program . . . . . . . . . . . . . . . . . 3-41

Example 3-23. Three Alternate Packing Methods for Avoiding Store Forwarding Difficulty . . . . 3-42

Example 3-24. Using Four Registers to Reduce Memory Spills and Simplify Result Passing. . . . . 3-43

Example 3-25. Stack Optimization Technique to Simplify Parameter Passing. . . . . . . . . . . . . . . . . . 3-43

Example 3-26. Base Line Code Sequence to Estimate Loop Overhead . . . . . . . . . . . . . . . . . . . . . . . . 3-45

Example 3-27. Loads Blocked by Stores of Unknown Address. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-47

Example 3-28. Code That Causes Cache Line Split . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-49

Example 3-29. Situations Showing Small Loads After Large Store . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-52

Example 3-30. Non-forwarding Example of Large Load After Small Store . . . . . . . . . . . . . . . . . . . . . 3-53

Example 3-31. A Non-forwarding Situation in Compiler Generated Code . . . . . . . . . . . . . . . . . . . . . . 3-53

Example 3-32. Two Ways to Avoid Non-forwarding Situation in Example 3-31. . . . . . . . . . . . . . . . 3-53

Example 3-33. Large and Small Load Stalls. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-54

Example 3-34. Loop-carried Dependence Chain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-56

Example 3-35. Rearranging a Data Structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-57

Example 3-36. Decomposing an Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-57

Example 3-37. Dynamic Stack Alignment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-59

Example 3-38. Aliasing Between Loads and Stores Across Loop Iterations. . . . . . . . . . . . . . . . . . . . 3-63

Example 3-39. Using Non-temporal Stores and 64-byte Bus Write Transactions . . . . . . . . . . . . . . 3-68

Example 3-40. On-temporal Stores and Partial Bus Write Transactions . . . . . . . . . . . . . . . . . . . . . . . 3-68

Example 3-41. Using DCU Hardware Prefetch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-71

Example 3-42. Avoid Causing DCU Hardware Prefetch to Fetch Un-needed Lines . . . . . . . . . . . . . 3-72

Example 3-43. Technique For Using L1 Hardware Prefetch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-73

Example 3-44. REP STOSD with Arbitrary Count Size and 4-Byte-Aligned Destination . . . . . . . . . 3-76

Example 3-45. Algorithm to Avoid Changing Rounding Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-82

xvii

CONTENTS

PAGE

Example 4-1. Identification of MMX Technology with CPUID. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-2

Example 4-2. Identification of SSE with CPUID. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-2

Example 4-3. Identification of SSE2 with cpuid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3

Example 4-4. Identification of SSE3 with CPUID . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3

Example 4-5. Identification of SSSE3 with cpuid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-4

Example 4-6. Identification of SSE4.1 with cpuid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-4

Example 4-7. Simple Four-Iteration Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-10

Example 4-8. Streaming SIMD Extensions Using Inlined Assembly Encoding . . . . . . . . . . . . . . . . . .4-11

Example 4-9. Simple Four-Iteration Loop Coded with Intrinsics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-12

Example 4-10. C++ Code Using the Vector Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-13

Example 4-11. Automatic Vectorization for a Simple Loop. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-14

Example 4-12. C Algorithm for 64-bit Data Alignment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-17

Example 4-14. SoA Data Structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-20

Example 4-15. AoS and SoA Code Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-20

Example 4-13. AoS Data Structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-20

Example 4-16. Hybrid SoA Data Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-22

Example 4-17. Pseudo-code Before Strip Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-23

Example 4-18. Strip Mined Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-24

Example 4-19. Loop Blocking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-25

Example 4-20. Emulation of Conditional Moves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-27

Example 5-1. Resetting Register Between __m64 and FP Data Types Code. . . . . . . . . . . . . . . . . . . 5-4

Example 5-2. FIR Processing Example in C language Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-5

Example 5-3. SSE2 and SSSE3 Implementation of FIR Processing Code . . . . . . . . . . . . . . . . . . . . . . . 5-5

Example 5-5. Signed Unpack Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-7

Example 5-4. Zero Extend 16-bit Values into 32 Bits Using Unsigned Unpack Instructions Code. 5-

Example 5-6. Interleaved Pack with Saturation Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-9

Example 5-7. Interleaved Pack without Saturation Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-10

Example 5-8. Unpacking Two Packed-word Sources in Non-interleaved Way Code. . . . . . . . . . . .5-12

Example 5-9. PEXTRW Instruction Code. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-13

Example 5-11. Repeated PINSRW Instruction Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-14

Example 5-10. PINSRW Instruction Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-14

Example 5-12. Non-Unit Stride Load/Store Using SSE4.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . .5-15

Example 5-13. Scatter and Gather Operations Using SSE4.1 Instructions . . . . . . . . . . . . . . . . . . . . . .5-15

Example 5-14. PMOVMSKB Instruction Code. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-16

Example 5-15. Broadcast a Word Across XMM, Using 2 SSE2 Instructions . . . . . . . . . . . . . . . . . . . . .5-17

Example 5-16. Swap/Reverse words in an XMM, Using 3 SSE2 Instructions . . . . . . . . . . . . . . . . . . .5-18

Example 5-17. Generating Constants. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-19

Example 5-18. Absolute Difference of Two Unsigned Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-21

Example 5-19. Absolute Difference of Signed Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-21

Example 5-20. Computing Absolute Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-22

Example 5-21. Basic C Implementation of RGBA to BGRA Conversion . . . . . . . . . . . . . . . . . . . . . . . . .5-22

Example 5-22. Color Pixel Format Conversion Using SSE2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-23

Example 5-23. Color Pixel Format Conversion Using SSSE3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-24

Example 5-24. Big-Endian to Little-Endian Conversion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-25

Example 5-25. Clipping to a Signed Range of Words [High, Low] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-26

Example 5-26. Clipping to an Arbitrary Signed Range [High, Low] . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-26

CONTENTS

xviii

PAGE

Example 5-28. Clipping to an Arbitrary Unsigned Range [High, Low] . . . . . . . . . . . . . . . . . . . . . . . . . . 5-27

Example 5-27. Simplified Clipping to an Arbitrary Signed Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-27

Example 5-29. Complex Multiply by a Constant. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-30

Example 5-30. Using PTEST to Separate Vectorizable and non-Vectorizable Loop Iterations . . . 5-31

Example 5-31. Using PTEST and Variable BLEND to Vectorize Heterogeneous Loops. . . . . . . . . . 5-32

Example 5-32. Baseline C Code for Mandelbrot Set Map Evaluation. . . . . . . . . . . . . . . . . . . . . . . . . . . 5-33

Example 5-33. Vectorized Mandelbrot Set Map Evaluation Using SSE4.1 Intrinsics . . . . . . . . . . . . 5-34

Example 5-34. A Large Load after a Series of Small Stores (Penalty) . . . . . . . . . . . . . . . . . . . . . . . . . 5-36

Example 5-36. A Series of Small Loads After a Large Store . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-37

Example 5-37. Eliminating Delay for a Series of Small Loads after a Large Store . . . . . . . . . . . . . . 5-37

Example 5-35. Accessing Data Without Delay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-37

Example 5-38. An Example of Video Processing with Cache Line Splits . . . . . . . . . . . . . . . . . . . . . . . 5-38

Example 5-39. Video Processing Using LDDQU to Avoid Cache Line Splits. . . . . . . . . . . . . . . . . . . . . 5-39

Example 5-40. Un-optimized Reverse Memory Copy in C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-41

Example 5-41. Using PSHUFB to Reverse Byte Ordering 16 Bytes at a Time. . . . . . . . . . . . . . . . . . 5-42

Example 5-42. PMOVSX/PMOVZX Work-around to Avoid False Dependency . . . . . . . . . . . . . . . . . . 5-45

Example 5-43. Table Look-up Operations in C Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-46

Example 5-44. Shift Techniques on Non-Vectorizable Table Look-up . . . . . . . . . . . . . . . . . . . . . . . . . 5-47

Example 5-45. PEXTRD Techniques on Non-Vectorizable Table Look-up . . . . . . . . . . . . . . . . . . . . . . 5-48

Example 6-1. Pseudocode for Horizontal (xyz, AoS) Computation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-6

Example 6-2. Pseudocode for Vertical (xxxx, yyyy, zzzz, SoA) Computation . . . . . . . . . . . . . . . . . . 6-6

Example 6-3. Swizzling Data Using SHUFPS, MOVLHPS, MOVHLPS . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-7

Example 6-4. Swizzling Data Using UNPCKxxx Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-8

Example 6-5. Deswizzling Single-Precision SIMD Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-9

Example 6-6. Deswizzling Data Using SIMD Integer Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-10

Example 6-7. Horizontal Add Using MOVHLPS/MOVLHPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-12

Example 6-8. Horizontal Add Using Intrinsics with MOVHLPS/MOVLHPS . . . . . . . . . . . . . . . . . . . . . 6-12

Example 6-9. Multiplication of Two Pair of Single-precision Complex Number . . . . . . . . . . . . . . . . 6-15

Example 6-10. Division of Two Pair of Single-precision Complex Numbers . . . . . . . . . . . . . . . . . . . . 6-16

Example 6-11. Double-Precision Complex Multiplication of Two Pairs . . . . . . . . . . . . . . . . . . . . . . . . . 6-17

Example 6-12. Double-Precision Complex Multiplication Using Scalar SSE2 . . . . . . . . . . . . . . . . . . . . 6-17

Example 6-13. Dot Product of Vector Length 4 Using SSE/SSE2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-19

Example 6-14. Dot Product of Vector Length 4 Using SSE3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-19

Example 6-15. Dot Product of Vector Length 4 Using SSE4.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-19

Example 6-16. Unrolled Implementation of Four Dot Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-20

Example 6-17. Normalization of an Array of Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-21

Example 6-18. Normalize (x, y, z) Components of an Array of Vectors Using SSE2. . . . . . . . . . . . . 6-22

Example 6-19. Normalize (x, y, z) Components of an Array of Vectors Using SSE4.1 . . . . . . . . . . . 6-23

Example 6-20. Data Organization in Memory for AOS Vector-Matrix Multiplication . . . . . . . . . . . . 6-24

Example 6-21. AOS Vector-Matrix Multiplication with HADDPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-24

Example 6-22. AOS Vector-Matrix Multiplication with DPPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-25

Example 6-23. Data Organization in Memory for SOA Vector-Matrix Multiplication . . . . . . . . . . . . 6-26

Example 6-24. Vector-Matrix Multiplication with Native SOA Data Layout . . . . . . . . . . . . . . . . . . . . 6-27

Example 7-1. Pseudo-code Using CLFLUSH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-13

Example 7-2. Populating an Array for Circular Pointer Chasing with Constant Stride . . . . . . . . . 7-15

Example 7-3. Prefetch Scheduling Distance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-18

Example 7-4. Using Prefetch Concatenation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-20

xix

CONTENTS

PAGE

Example 7-5. Concatenation and Unrolling the Last Iteration of Inner Loop . . . . . . . . . . . . . . . . . .7-20

Example 7-6. Data Access of a 3D Geometry Engine without Strip-mining . . . . . . . . . . . . . . . . . . .7-26

Example 7-7. Data Access of a 3D Geometry Engine with Strip-mining. . . . . . . . . . . . . . . . . . . . . . .7-26

Example 7-8. Using HW Prefetch to Improve Read-Once Memory Traffic. . . . . . . . . . . . . . . . . . . . .7-28

Example 7-9. Basic Algorithm of a Simple Memory Copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7-33

Example 7-10. A Memory Copy Routine Using Software Prefetch. . . . . . . . . . . . . . . . . . . . . . . . . . . . .7-34

Example 7-11. Memory Copy Using Hardware Prefetch and Bus Segmentation . . . . . . . . . . . . . . . .7-36

Example 8-1. Serial Execution of Producer and Consumer Work Items . . . . . . . . . . . . . . . . . . . . . . . . 8-6

Example 8-2. Basic Structure of Implementing Producer Consumer Threads . . . . . . . . . . . . . . . . . . 8-7

Example 8-3. Thread Function for an Interlaced Producer Consumer Model . . . . . . . . . . . . . . . . . . . 8-9

Example 8-4. Spin-wait Loop and PAUSE Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8-17

Example 8-5. Coding Pitfall using Spin Wait Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8-20

Example 8-6. Placement of Synchronization and Regular Variables . . . . . . . . . . . . . . . . . . . . . . . . . .8-22

Example 8-7. Declaring Synchronization Variables without Sharing a Cache Line . . . . . . . . . . . . .8-22

Example 8-8. Batched Implementation of the Producer Consumer Threads . . . . . . . . . . . . . . . . . .8-29

Example 8-9. Parallel Memory Initialization Technique Using OpenMP and NUMA . . . . . . . . . . . . .8-34

Example 10-1. A Hash Function Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10-5

Example 10-2. Hash Function Using CRC32. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10-6

Example 10-3. Strlen() Using General-Purpose Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10-9

Example 10-4. Sub-optimal PCMPISTRI Implementation of EOS handling . . . . . . . . . . . . . . . . . . . . 10-11

Example 10-5. Strlen() Using PCMPISTRI without Loop-Carry Dependency. . . . . . . . . . . . . . . . . . . 10-12

Example 10-6. WordCnt() Using C and Byte-Scanning Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-13

Example 10-7. WordCnt() Using PCMPISTRM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-15

Example 10-8. KMP Substring Search in C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-17

Example 10-9. Brute-Force Substring Search Using PCMPISTRI Intrinsic . . . . . . . . . . . . . . . . . . . . . 10-19

Example 10-10.Substring Search Using PCMPISTRI and KMP Overlap Table . . . . . . . . . . . . . . . . . . 10-22

Example 10-11.I Equivalent Strtok_s() Using PCMPISTRI Intrinsic . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-26

Example 10-12.I Equivalent Strupr() Using PCMPISTRM Intrinsic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-29

Example 10-13.UTF16 VerStrlen() Using C and Table LookupTechnique . . . . . . . . . . . . . . . . . . . . . 10-31

Example 10-14.Assembly Listings of UTF16 VerStrlen() Using PCMPISTRI . . . . . . . . . . . . . . . . . . . 10-32

Example 10-15.Intrinsic Listings of UTF16 VerStrlen() Using PCMPISTRI . . . . . . . . . . . . . . . . . . . . . 10-35

Example 10-16.Replacement String Library Strcmp Using SSE4.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-38

Example 10-17.High-level flow of Character Subset Validation for String Conversion. . . . . . . . . 10-40

Example 10-18.Intrinsic Listings of atol() Replacement Using PCMPISTRI. . . . . . . . . . . . . . . . . . . . . 10-41

Example 10-19.Auxiliary Routines and Data Constants Used in sse4i_atol() listing. . . . . . . . . . . . 10-44

Example 12-1. Instruction Pairing and Alignment to Optimize Decode Throughput on Intel® Atom™ Mi-

croarchitecture 12-5

Example 12-2. Alternative to Prevent AGU and Execution Unit Dependency . . . . . . . . . . . . . . . . . .12-8

Example 12-3. Pipeling Instruction Execution in Integer Computation . . . . . . . . . . . . . . . . . . . . . . . . .12-9

Example 12-4. Memory Copy of 64-byte . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12-14

Example 12-5. Examples of Dependent Multiply and Add Computation . . . . . . . . . . . . . . . . . . . . . . 12-16

Example 12-6. Instruction Pointer Query Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12-17

Example 12-8. Auto-Generated Code of Storing Absolutes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-8

Example 12-9. Changes Signs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-8

Example 12-7. Storing Absolute Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-8

Example 12-11.Data Conversion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-9

Example 12-10.Auto-Generated Code of Sign Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-9

剩余603页未读，继续阅读

lch21

粉丝: 2
资源: 1

Intel 64位汇编优化手册：加速程序性能

IA-64架构下的64位汇编语言编程工具与技巧

64位英特尔汇编语言中的位操作

深入掌握64位汇编语言与nasm编译器教程

intel汇编语言 汇编语言学习文档

AT&T汇编查询文档

8086汇编帮助文档.zip

AT&T汇编语言文档

intel.rar_Intel 汇编 语言_intel汇编_汇编

intel指令集（适合于从底层汇编优化开发）

64位汇编编程学习工具和资料

最新资源

intel汇编语言汇编语言学习文档

intel.rar_Intel 汇编语言_intel汇编_汇编