Intel Corp.
This paper teaches the programmer how to get started with the Streaming SIMD Extensions and Streaming SIMD Extensions 2 instruction sets that are available for the Intel® Pentium 4 Processor. In this paper, we introduce these techniques by presenting the environment requirements, why to use SSE/SSE2 instructions, and examples that use this type of coding technique [3,395].
Single-instruction multiple-data execution is the concept of performing multiple computations with a single instruction in parallel. This technology, introduced with MMX technology, allows SIMD computations to be performed using the 64-bit MMX registers.
Streaming SIMD extensions were introduced in the P6 microarchitecture for the Intel Pentium III processor, extended MMX technology, and allows SIMD computations to be performed on four packed single-precision floating-point data elements simultaneously. These computations are performed concurrently 128 bits at a time.
SSE2 were introduced in the Intel NetBurst architecture. SSE2 allowed the ability to perform more computations in parallel, and extended those instructions introduced in MMX and SSE. Notably, SSE2 introduces SIMD computations on two double-precision floating-point data elements.
© 2004 Intel Corp.
If you're interested in this topic, these articles may be helpful:
 | Intel Itanium Processor Family reference guide: IA-32 Execution Layer The Intel Itanium processor family has been architected to provide ind... |
| Intel® Math Kernel Library for Windows The flagship for HPC software, Intel® MKL is a set of highly optim... |
| Intel Fortran Compiler 8.1 for Linux Intel Corp Reduce time-to-market, optimize your code for Intel®... |
| Branch and loop reorganization to prevent misdirects by Jeff Andrews, application engineer, Intel Corp. Proven technique... |
| Intel Math Kernel Library 7.0 extends benefits by Matt Gillespie, technical author and editor. Intel Corp. The Int... |
Related Jobs:
