Professional Assembly Language

Chapter 17. Using Advanced IA-32 Features

The Intel Pentium family of processors provides advanced mathematical processing capabilities to the IA-32 instruction set. If you are working with applications that require a lot of mathematical processing, such as audio and video processing, utilizing the advanced mathematical features can greatly increase the performance of your application.

This chapter describes the Single Instruction Multiple Data (SIMD) family of instructions used on Pentium processors. First, a review of the main features of SIMD is presented. Following that, how to use the MMX architecture instructions is shown. Next, the SSE features are discussed, along with a demonstration of how to use SSE to process multiple data elements in a single instruction. Finally, the SSE2 features are shown, along with an example of using them in an assembly language program.

A Brief Review of SIMD

Chapters 2, "The IA-32 Platform," and 7, "Using Numbers," described the various features and data types found in the IA-32 SIMD architecture. To recap, the IA-32 SIMD architecture currently consists of four technologies:

• Multimedia Extensions (MMX)

• Streaming SIMD Extensions (SSE)

• Streaming SIMD Extensions Second Implementation (SSE2)

• Streaming SIMD Extension Third Implementation (SSE3)

As discussed in Chapter 2, different Intel Pentium processors support different versions of the SIMD architecture. The Pentium MMX and Pentium II processors support MMX. The Pentium III processor supports both MMX ...