使用“英特尔集成性能原件(英特尔 IPP)Windows 版”,可构建随选即用的功能并提高应用程序性能。它利用跨多个平台的函数库,这些函数库中的函数均针对英特尔酷睿双核处理器的多核心功能进行过优化。您还可以使用各种预先构建的函数与编,涉及领域包括:音频、视频及图像处理;密码技术;语音识别;以及矩阵、矢量数学及信号处理。
Intel® Integrated Performance Primitives (Intel® IPP) is an extensive library of multicore-ready, highly optimized software functions for multimedia data processing, and communications applications.Intel IPP is one of the Intel® Performance Libraries, which provide optimizing software building blocks to complement Intel’s optimizing compilers and performance optimization tools. Intel IPP is available as a component of the Intel® Compiler Professional Editions and Compiler Suite Editions for a more complete and cost-effective solution or as a standalone product.
功能 多核处理器支持 图 1. 多核处理器使得多线程软件应用真正实现了并行执行
“英特尔® 集成性能原件”(英特尔® IPP)5.1 全面支持当今的各种多核心计算平台:
- 多核心优化函数:“矢量与统计”数学、“信号过滤”、“傅立叶变换”、“图像/JPEG 压缩与颜色转换”等领域的许多关键函数的内部都使用 OpenMP* 进行过多线程处理,可以在多核心系统上实现性能。
- 多核心优化代码示例:许多“英特尔 IPP”代码示例都进行过多线程处理,以说明在视频编码与解码等应用程序中如何有效地使用“英特尔 IPP”函数。
- 真正线程安全的函数:所有的“英特尔 IPP”函数都完够确保线程安全,简化了集成到多线程应用程序中的工作。
如需有关线程技术与“英特尔 IPP”函数的详细信息,请参阅线程技术的常见问题 (FAQ) 页面。
- 性能优化的函数
“英特尔 IPP 5.1”中添加了一些新的优化技术,可以在的处理器(如英特尔® 酷睿™ 双核处理器与英特尔® 奔腾® D 处理器)上实现性能。
使用众多技术领域中广泛的经过高度优化的函数,消除应用程序中可能出现的任何性能瓶颈:
- 视频编码:DV、MPEG-2、MPEG-4、H.263 以及 MPEG-4 Part 10 (H.264) 编的关键算法组件。“图 2”显示 H.264 编流程中适合使用“英特尔 IPP”视频编码组件(用蓝色方框表示)的地方。这些函数包括:
- 运动补偿
- 运动估计
- 改进的离散余弦变换
- 量化与反量化
- 熵编码
“视频”与“音频”代码示例演示如何使用“英特尔 IPP”函数实现编。
- 图像处理:包含许多内置的图像处理函数,其中有:
- 计算机视觉:提供视频数据处理功能,包括:
- 颜色转换:采用以下技术改善图像颜色与颜色效果之间的转换效率:
- 字符串处理:使用“字符串处理”(查找、插入、删除、比较)与“常规表达式”,将文本数据库管理、搜索与检索或文档索引处理功能集成到应用程序中。
备注:“英特尔® PCA 处理器”上不提供。 - JPEG 编码:JPEG、JPEG 2000 以及“动态 JPEG”编的关键算法组件。“图 3”显示 JPEG 与 JPEG 2000 编流程中适合使用“英特尔 IPP”的 JPEG 编码组件(用蓝色方框表示)的地方。
备注:“英特尔® PCA 处理器”上不提供。 - 语音编码:包含针对以下用途的函数:
它还包含用于“回声消除”与“通用语音类别”的语音函数。 - 信号处理:包含针对以下用途的信号处理功能:
备注:“英特尔® PCA 处理器”上不提供。 - 数据压缩:采用以下技术地压缩多媒体数据:
- 音频编码:MP3、AAC 以及 AC3 编的关键算法组件。“图 4”显示 AAC 编流程中适合使用“英特尔 IPP”的 JPEG 编码组件(用蓝色方框表示)的地方。这些函数包括:
- 哈夫曼编码
- 光谱数据预量化
- 改进的离散余弦变换
- 区块过滤
- 频域预测
- 光谱带复制
- 快速傅立叶变换
- 语音识别:通过采用以下技术,使应用程序支持语音识别、“IP 语音传输”以及语音注解功能,从而实现口授和语音命令:
备注:“英特尔® PCA 处理器”上不提供。 - 矢量/矩阵运算:包含针对以下用途的实用矩阵与矢量函数:
备注:“英特尔® PCA 处理器”上不提供。 - 密码技术:采用以下加密算法在应用程序中编写安全功能代码:
备注:“英特尔® PCA 处理器”上不提供。
此版本的新增功能 - 新推出 Mac OS* 版!
获取适用于 Mac OS* 应用程序的高性能数据处理与多媒体函数库 - 所有这些使用的都是您熟悉的“英特尔 IPP API”。 - 支持多核处理器
使用能够确保线程安全的函数以及多线程函数,让应用程序无需修改便可充分发挥多核处理器的优势。 - 新的 CPU 性能优化技术
通过使用更新的、特定于英特尔® 酷睿™ 双核与英特尔® 酷睿™ 单核处理器微体系结构的优化技术,充分发挥这些新处理器的优势。 - 更加丰富的视频编码示例
使用新的代码示例(演示对附加的 H.264 编档案与 CABAC 编码的支持),更快地开发*的视频应用。 - 更加丰富的语音编码示例:
立即使用新的 AMRWB+ 代码示例开发协作应用。
兼容性 - 操作系统
现已支持 Mac OS*!借助“英特尔® 集成性能原件 5.1 Mac OS* 版”,只需使用 Windows* 版与 Linux* 版所有的相同函数库 API 与函数,便可以轻松将应用程序移植到 Mac OS。 - 开发环境
“英特尔 IPP”非常易于使用,且可轻松集成到主流的开发工具与环境中,如 Microsoft Visual Studio*、Xcode*、Eclipse*、GCC 以及“英特尔® C++ 编译器”。
处理器“英特尔 IPP”函数的背后是*一致的 API,这些函数针对广泛的 32 位与 64 位微处理器进行过高度优化:
- 英特尔® 酷睿™ 双核处理器与英特尔® 奔腾® D 多核处理器
- 英特尔® 至强® 处理器
- 含“英特尔® EM64T”的处理器,包括 64 位英特尔至强处理器、英特尔® 奔腾® D 处理器以及英特尔® 奔腾® 处理器版
- 英特尔® 奔腾® 4 与英特尔® 奔腾® M 处理器
- 基于“英特尔 XScale®”技术的处理器,包括英特尔® IXP4xx 处理器与支持“英特尔® 无线 MMX™”技术的英特尔® PXA27x 应用处理器。
- 英特尔® 安腾® 2 处理器
Intel IPP is validated for use with multiple generations of Intel and compatible AMD* processors, and is backed by world-class support through the Intel® Premier Support program, and by developer community forums.
Features
Multicore Processor Support
Intel IPP 6.0 fully supports today’s multicore computing platforms:
- Multicore Optimized, Threaded Functions: Over 1700 critical functions for Matrix and Vector mathematics, Signal/Image Filtering and Convolutions, Image/JPEG Compression, Color Conversion and Computer Vision are internally threaded to automatically maximize performance on multi-core systems.
- Multicore Optimized Code Samples: Many of the Intel IPP code samples are threaded to illustrate the effective use of Intel IPP functions in applications such as video encoding and decoding.
- Fully Thread-Safe Functions: All Intel IPP functions are fully thread-safe, simplifying integration into threaded applications.
To learn more about threading and Intel IPP functions visit our threading Frequently Asked Questions (FAQ) page.
Performance-Optimized Functions
Intel IPP functions are designed to deliver performance beyond what optimized compilers alone can deliver by matching the function algorithms to low-level optimizations based on the processor’s available features such as Streaming SIMD Extensions (SSE, SSE2, SSE3, SSSE3, SSE4, and SSE4.1) and other optimized instruction sets.
Video Coding: Key algorithmic components for DV25/50/100, MPEG-2, MPEG-4, H.263, and MPEG-4 Part 10 (H.264) codecs. Figure 3 shows where Intel IPP video coding components (represented by the blue boxes) fit into the H.264 codec process flow. Functions include:
- Motion Compensation
- Motion Estimation
- Modified Discrete Cosine Transforms
- Quantization and Inverse Quantization
- Entropy Coding
Image and 2D Signal Processing: Intel IPP is the premier library of image and 2D signal processing algorithms, and includes a rich selection of algorithms operating on images and regions within images (ROIs):
Transforms: Wavelet Fourier (FFT/DFT, real/complex) Windowing (Hamming, Bartlett) Discrete Cosine (DCT)
Filtering Functions: General linear filtering Convolution/Deconvolution (LR and FFT) Box, Min, Max, Median Wiener Filters Fixed Filters (Prewitt, Sobel, Laplace, Gauss, Scharr, Roberts) Sharpening/Hipass/Lowpass Filters
Geometric Transforms: Resize, Mirror, Rotate, Shear Affine transforms Perspective transformations Bilinear warping Coordinate remapping | Image Statistics: Sum, Integral, Tilted Integral Mean, Min, Max, Histogram, StDev Image Moments Image Norms (L1, L2, infinity) Image Quality Index calculation Proximity Measures (Cross-correlations, Square Distance) Threshold/Compare Operations
Image Arithmetic/Logic Operations: Alpha composition Arithmetic operations (add/sub/mul/div/sqrt/sqr/ln/exp/abs) Logical operations (AND, OR, XOR, Shift, NOT)
Image Data Exchange/Initialization: Copy/Set/Transpose Channel swapping Jaehne/Ramp/Zigzag initialization Memory allocation for multiple image types |
Computer Vision: Intel IPP includes optimized functions for many key computer-vision operations, for applications in security, machine control, media management, media annotation and more:
Feature Detection (Corner, Canny Edge Detection) Distance Transforms Image Gradients Flood Filling Motion Templates generation Optical Flow calculation (Lucas-Kanade) | Pattern Recognition (Haar classifiers) Pyramid functions (Gaussian/Laplacian pyramids) Universal Pyramid functions Camera Calibration 3D Reconstruction |
Intel IPP based optimization is automatically included in the popular OpenCV open-source computer-vision library, for enhanced performance on real-time tasks, and Intel IPP was a key software component in the winner of the 2005 DARPA Grand Challenge.
Color Conversion: Today’s explosion of digital media in multiple formats brings the need to convert digital media among different color representations. Intel IPP provides a rich set of optimized color-conversion routines on 32/24/16-bit-per-pixel formats:
Color Model Conversion: RGB, YUV, YCbCr, BGR, CbYCr, HSV, LUV, Lab, YCC, HLS, SBGR, YCoCg, YCCK, XYZ, CMYK
Color Format Conversions: YCbCr422, YCbCr420, YCbCr411, CbYCr422, BGR565, BGR555, BGR565Dither | Lookup Table Conversions (Linear/Cublic/Palette) Color to Greyscale Conversions (Fixed/Custom coefficients) Image Bit Resolution Reductions Color Twist Conversions (integer/float pixel values) Gamma Corrections (Forward/Backward) |
[SPAN]
String Processing: Build optimized text database management, search and retrieval, or document indexing processing into your applications using Intel IPP’s optimized string operations.
Substring substitution/insertion String concatenation/splitting Upper/lower case conversions | String/Substring matching Regular Expression matching Hash value calculation |
JPEG Coding: Key algorithmic components for JPEG, JPEG 2000, and Motion JPEG codecs. Figure 4 shows where Intel IPP JPEG coding components (represented by the blue boxes) fit into the JPEG and JPEG 2000 codec process flow.
Speech Coding: Intel IPP includes a comprehensive set of routines supporting the following speech codecs/functions:
G.722.1 G.722 Sub-Band ADPCM G.723.1 G.726 G.728 Echo Cancellation | G.729 GSM-AMR AMR-Wideband GSM Full Rate Companding |
The freely-downloadable Intel IPP Universal Speech Class (USC) code samples illustrate the use of the low-level Intel IPP functions to build speech codecs.
Signal Processing: Includes signal processing features for the following:
Filtering and Convolution: Finite Impulse Response (FIR) Infinite Impulse Response (IIR) Median filtering Cyclic convolution Auto/Cross-correlation
Transforms: Fourier (FFT, DFT, Goertzel) Discrete Cosine Transforms (DCT) Hilbert Transforms Wavelet Transforms (fixed/custom filters) Power spectrum calculation
Windowing/Sampling: Upsampling/Downsampling Windowing (Bartlett/Blackman/Hamming/Hann/Kaiser) | Array/Signal Initialization/Manipulation: Move/Copy/Set/Zero Tone/Triangle/Ramp/Jaehne Generation Random vector generation (Uniform/Gaussian) Array allocation Real/Complex conversion Polar/Cartesian conversion
Array/Signal statistics: Sum/Max/Min/Mean/StdDev/Norm Dot products Thresholding Viterbi decoding
Array Arithmetic/Logic Operations: Arithmetic operations (add/sub/mul/div/sqrt/sqr/ln/exp/abs) Logical operations (AND, OR, XOR, Shift, NOT) Array sorting Magnitude/Phase |
Data Compression: In addition to video, audio and image compression with codecs, Intel IPP provides functions for lossless compression methods, such as those used in the popular “zlib” (inflate and deflate) and “libbzip2” libraries.
Burrows-Wheeler-Transform techniques: Burrows-Wheeler Transform (BWT) Generalized Interval Transform Move-to-Front (MTF) Run-length encoding (RLE) | Entropy coding: Huffman coding Variable-Length Coding (VLC)
Dictionary-based Compression: LZSS encode/decode LZ77 encode/decode |
Audio Coding: Key algorithmic components for MP3,and AAC codecs. Figure 5 shows where Intel IPP JPEG coding components (represented by the blue boxes) fit into the AAC codec process flow. Functions include:
- Huffman Coding
- Spectral Data Pre-Quantization
- Modified Discrete Cosine Transforms
- Block Filtering
- Frequency Domain Prediction
- Spectral Band Replication
- Fast Fourier Transforms
The Video and Audio code samples illustrate sample codec implementations using Intel IPP functions.[SPAN]
Speech Recognition: Build advanced speech recognition, Voice-over-IP, and voice annotation capabilities in applications, using Intel IPP’s broad range of speech-recognition capabilities:
Feature Processing Model Evaluation Model Estimation Model Adaptation Vector Quantization | Acoustic Echo Cancellation (AEC) Polyphase Resampling Advanced Aurora Functions Ephraim-Malah Noise Suppression Voice Activity Detection |
Vector/Matrix Operations: Intel IPP contains a rich set of matrix and vector operations for a wide variety of applications, including physics modeling and 3D transform/lighting calculations.
Matrix algebra: Eigenvalue/eigenvector calculation Least Squares (QR decompositions/back-sub) Linear Systems (LU/Cholesky) Region-of-Interest (ROI) extraction Fast copy of vectors/matrices | Vector algebra: Dot products L2 norm calculation “saxpy” (ax + y) operations Linear combination (ax + by) Power/root functions Exponential/Logarithmic/Erf/Erfc functions Trigonometric/Hyperbolic functions Polar/Cartesian conversion |
For applications requiring high-performance linear algebra operations on very large data sets, the Intel® Math Kernel Library may also be of interest.
Cryptography: Use Intel IPP to quickly build robust, high-performance cryptographic modules and applications. Below are some of the many cryptographic building blocks included in Intel IPP’s cryptography functions.
Symmetric Ciphers: Block Ciphers (AES/Rijndael, DES, Triple DES, Blowfish, Twofish) Stream Ciphers (ARCFour)
One-way Hashing: Generalized Hashing (MD5, SHA1-512) Mask Generation (MD5, SHA1-512)
Data Authentication: Keyed Hash (HMAC-MD5, HMAC-SHA1-512) Data Authentication Functions (DES, TDES, Rijndael, Blowfish, Twofish) | Asymmetric Cryptography: Elliptic curve cryptography (GF(p) and GF(2m) RSA algorithm (RSA-OAEP, RSA-SSA) Discrete-Logarithm Cryptography Big-Number arithmetic Montgomery reduction Pseudo-random number generation Prime number generation |
Intel IPP’s cryptographic functions have been validated according to the Cryptographic Algorithm Validation Program (CAVP).
Name | Standards/Certificate |
Discrete Logarithm (Digital Signature Standard) | FIPS 186-2 / Cert 190 |
Integer Factoring (Digital Signature Standard) | FIPS 186-2, ANSI X9.31-1998 / Cert 181 |
Eliptic Curve Digital Signature Algorithm (ECDSA) | FIPS 186-2, ANSI X9.62-1998 / Cert 40 |
Random Number Generator | FIPS 186-2 / Cert 245 |
Note: To access the Cryptography Library, you must request access from Intel.
Ray-Tracing and Rendering: Core operations used in ray-tracing, realistic image rendering, and physics applications:
Bounding-box calculations Object-ray intersections Shadow/Reflection calculations |
Data Integrity: Error correcting codes are vitally important to preserving the integrity of data in transmission, storage and encoding. For example transmission lines can be unreliable and introduce data errors, spurious signals can occur when saving data to a compact disk, and errors can occur when reading bar codes. Using error correcting codes like Reed-Solomon is a good way to correct these errors.
System Requirements
Please refer to the section below for installation requirements and system requirements that match your application’s target platform.[SPAN]
Application Target Platforms
32-bit Intel® architecture Platforms | Intel® Core i7™ processor family | Intel® Atom™ processor family | Intel® Core™ 2 processor family(including 45nm processors) | Intel Core processor family | Intel® Pentium® M processors | Intel ultra- processors (A100, A110, and 45 nm processors) | Intel Pentium D processor | Intel® Pentium® 4 processor | Intel® Xeon® processors(3000, 5000, 7000 series) | Other compatible processors such as AMD Athlon* and Opteron* processors |
|
64-bit Intel 64 architecture-based platforms | Intel® Core i7™ processor family | Intel® Atom™ processor family | Intel Core 2 processor family (including 45nm processors) | Intel ultra- processors (45 nm processors) | Intel Pentium D processor | Intel Pentium 4 processor | Intel Xeon processors(3000, 5000, 7000 series) | Other compatible processors such as AMD Athlon and Opteron processors |
|
64-bit IA-64 architecture-based platforms | Intel® Itanium® processors |
|
32-bit IA-32 Architecture-Based and Compatible Platforms
(Intel Core 2 processor family, Intel Core processor family, Intel Pentium processors, and compatible AMD processors)
| OS Version | Supported Compilers |
Microsoft Windows | Microsoft Windows Vista*
Microsoft Windows* XP
Microsoft Windows Server* 2008
Microsoft Windows Server* 2003 | Intel® C++ Compiler for Windows* OS version 10.1 and 11.0
Microsoft Visual Studio* 2008
Microsoft Visual Studio* 2005
Microsoft Visual C++* .NET 2003
Microsoft Windows* Software Development Kit for Microsoft Windows Vista* |
Linux | Linux system with glibc 2.2.4, 2.2.5, 2.2.93, 2.3.2 or 2.3.3 and the 2.4.X or 2.6.X Linux kernel as represented by the following distributions. Note: Not all distributions listed are validated and not all distributions are listed. - Red Hat Enterprise Linux* OS 2.1, 3, or 4
- SUSE* Linux 8.2, 9.1
- SUSE Linux Enterprise Server* 8 or 9
- Debian* 4.0r1
- Ubuntu* 7.10
- Asianux* Server 3.0
- Red Flag* 5.0
| Intel® C++ Compiler version 10.1 and 11.0 for Linux * for IA-32 architecture
Linux Developer tools component installed, including gcc, g++ and related tools |
Mac OS X | 10.4.3 or later | Intel C++ Compiler 10.1 and 11.0 for Mac OS X, including Professional Edition
GNU Compiler Collection 4.0 or later |
64-bit Intel 64 Architecture-Based Platforms
(Intel Core 2 processor family, Intel Pentium D processors, Intel Xeon processors, and processors from AMD compatible with Intel 64 architecture)
| OS Version | Supported Compilers |
Microsoft Windows | Microsoft Windows Vista*
Microsoft Windows* XP
Microsoft Windows Server* 2008
Microsoft Windows Server* 2003 | Intel® C++ Compiler for Windows* OS version 10.1 and 11.0 for processors based on Intel® 64 architecture
Microsoft Visual Studio* 2008
Microsoft Visual Studio* 2005
Microsoft Platform SDK, Version 3790.1830 (April 2005)
Microsoft Platform SDK R2, Version 3790.2075 (March 2006) |
Linux | Linux system with glibc 2.2.4, 2.2.5, 2.2.93, 2.3.2 or 2.3.3 and the 2.4.X or 2.6.X Linux kernel as represented by the following distributions. Note: Not all distributions listed are validated and not all distributions are listed. - Red Hat Enterprise Linux* OS 3 or 4
- SUSE Linux Enterprise Server* 9
- Debian* 4.0r1
- Ubuntu* 7.10
- Asianux* Server 3.0
- Red Flag* 5.0
| Intel C++ Compiler 10.1 or 11.0 for Linux
Linux Developer Tools installed, including gcc, g++, and related tools |
Mac OS X | 10.4.6 or later | Intel C++ Compiler 10.1 and 11.0 for Mac OS X
GNU Compiler Collection 4.0 or later |
[SPAN]
64-bit IA-64 Architecture-Based Platforms
(Intel® Itanium® processors)
| OS Version | Supported Compilers |
Microsoft Windows | Microsoft Windows Server 2008
Microsoft Windows Server 2003 | Intel® C++ Compiler for Windows* OS version 10.1 and 11.0 for Intel Itanium Processors
Microsoft Platform SDK, Version 3790.1830 (April 2005)
Microsoft Platform SDK R2, Version 3790.2075 (March 2006) |
Linux | Linux system with glibc 2.2.4, 2.2.5, 2.2.93, 2.3.2 or 2.3.3 and the 2.4.X or 2.6.X Linux kernel as represented by the following distributions. Note: Not all distributions listed are validated and not all distributions are listed. - Red Hat Enterprise Linux* OS 2.1, 3, or 4
- SUSE Linux Enterprise Server* 8 or 9
- Debian* 4.0r1
- Ubuntu* 7.10
- Asianux* Server 3.0
- Red Flag* 5.0
| Intel C++ Compiler 10.1 and 11.0 for Linux
Linux Developer Tools installed, including gcc, g++, and related tools |
Intel Atom Architecture-based Platforms
| OS Version | Supported Compilers |
| | |
Linux | Linux system with glibc 2.2.4, 2.2.5, 2.2.93, 2.3.2 or 2.3.3 and the 2.4.X or 2.6.X Linux kernel as represented by the following distributions.
Note: Not all distributions listed are validated and not all distributions are listed. - Ubuntu* 7.10
- Red Flag* 5.0
| Intel® C++ Compiler version 10.1 & 11.0 for Linux* OS for IA-32 processors
Linux Developer tools component installed, including gcc, g++ and related tools |
Installation Requirements
| Supported Software / Hardware |
Operating Systems | Microsoft Windows: - Windows Vista
- Windows XP, SP1 or SP2
- Windows Server 2003, SP1 or SP2
Linux:
- Ubuntu 6.04, 7.10
- Debian 3.1r5, 4.0
- Red Hat Enterprise Linux* 3, 4 and 5
- Red Hat Fedora Core 4, 5
- SUSE Linux Enterprise Server* 9 or 10
- Red Flag DC Server 5.0
- Mandriva 10.1
Mac OS X:
|
Processor Requirements | IA-32 architecture-based and compatible platforms: - Intel® Pentium® III processor or later, 400MHz or faster
- Any Intel Core family processor
- Any Intel Xeon processor
- Compatible AMD processors
Intel 64 architecture-based and compatible platforms:
- Any Intel Core family processor
- Any Intel Xeon processor
- Any Intel Pentium D processor
- Compatible AMD processors
Intel Itanium processor-based platforms:
- Any Intel Itanium processor
Intel Atom processor-based platforms:
- Any Intel Atom family processor
|
Minimum Disk Space for Install | 900 MB plus 300 MB of temporary space during installation for each Application Target Platform |