Streaming SIMD Extensions 2

Streaming SIMD Extensions 2
Name	Streaming SIMD Extensions 2
Developer	Intel Corporation
Introduced	2001
Architecture	x86
Extensions	SSE
Predecessor	MMX
Successor	SSE3

Streaming SIMD Extensions 2

Overview

Streaming SIMD Extensions 2 is an extension to the x86 architecture introduced by Intel Corporation in 2001 alongside processors such as the Pentium 4 and contemporaneous with developments from Advanced Micro Devices, IBM, VIA Technologies, Transmeta, and ARM Holdings in the early 2000s to improve multimedia and numeric performance across platforms used by Microsoft, Apple Inc., Sony Corporation, Nintendo, and Sun Microsystems.

Technical Features

SSE2 expanded the existing Streaming SIMD Extensions family to support 128-bit registers compatible with the x87, MMX (instruction set), SSE, x86-64 register sets and added integer and floating-point operations used in workloads for Adobe Systems, Autodesk, Oracle Corporation, Siemens, and SAP SE to accelerate tasks typical for the Intel Pentium M, Intel Xeon, AMD Athlon 64, VIA C7 families.

Instruction Set Extensions

The instruction extensions introduced by SSE2 included packed double-precision floating-point operations, integer SIMD operations, and new moves and conversions used by compilers from GNU Project, Microsoft Visual Studio, Intel C++ Compiler, Borland, and LLVM to target optimizations relevant to applications from NVIDIA, ATI Technologies, Mathematica, MATLAB, and Blender.

Hardware and Implementation

SSE2 was implemented in hardware microarchitectures from Intel such as the NetBurst microarchitecture, Core microarchitecture, and later Nehalem, and in implementations from Advanced Micro Devices in designs like the K8 microarchitecture and AMD Opteron; chipset and motherboard vendors including ASUS, Gigabyte Technology, MSI (company), Intel 810 chipset, and VIA KT266 integrated support into consumer and server platforms.

Software Support and Compilers

Compiler support for SSE2 arrived in versions of GCC, Microsoft Visual C++, Intel Compiler, Clang/LLVM, Watcom, and Turbo C++ enabling automatic vectorization and intrinsics used by libraries from Intel Math Kernel Library, OpenBLAS, FFTW, libjpeg, libpng, and applications like Adobe Photoshop, Microsoft Office, AutoCAD, and Final Cut Pro.

Performance and Use Cases

SSE2 improved performance for double-precision scientific computing, image processing, audio codecs, cryptographic routines, and physics engines targeted by developers at Google, Facebook, Amazon, IBM Research, and NASA; real-world benefits were measurable in benchmarks produced by SPECint, SPECfp, Linpack, SiSoft Sandra, and workload suites from PCMark.

Compatibility and Legacy Issues

Adoption of SSE2 introduced compatibility considerations for operating systems and boot environments including Microsoft Windows XP, Windows Server 2003, Red Hat Enterprise Linux, Debian, Ubuntu, FreeBSD, and firmware from Phoenix Technologies and AMI; later transitions to x86-64 and vector extensions like SSE3, SSE4, AVX, AVX2, and AVX-512 built upon and in some cases deprecated assumptions from SSE2, creating migration paths used by vendors such as Canonical, Red Hat, Microsoft, Apple, and Oracle Corporation.

Category:Instruction set extensions