AMD recognizes the GNU toolset as a critical part of the software development ecosystem. Since the year 2000, AMD has been actively contributing to the evolution of the GNU toolset, particularly the GCC  project, the glibc  project, and the binutils project which includes GDB. Throughout the development, validation and delivery of GCC version 3.x beginning in 2001 and version 4.x beginning in 2005, AMD has contributed significant GCC code generation and reliability improvements for all x86 platforms. We feel strongly about our role, as a part of the greater open source community, to further drive the quality and adoption of GCC and other components of the GNU toolset.

» Downloads
» Documentation

GCC, available via Linux distributions (e.g. Red Hat and SUSE) and from the Free Software Foundation (FSF), supports the latest AMD processor-based platforms.  GCC 4.3.0 includes improvements for optimized code generation for AMD platforms composed of AMD “Istanbul” Family 10h processors (e.g. AMD Phenom™ X4, AMD Phenom™ X3, and Quad-Core AMD Opteron™ processors) and even supports proposed instruction set extensions for future AMD products.

The AMD contributions to GNU projects help enable developers using GCC to best leverage today’s architectures and to help future-proof current software products.  In addition, you can evaluate proposed instruction sets.  The following are some of the new features available in GCC 4.3.0 that can be leveraged for optimal execution on AMD processor-based platforms.

• GCC 4.3.0 with binutils 2.18 supports AMD “Istanbul” processor-based platforms.  Optimal code generation can be obtained with the optimization flag: -march=amdfam10.
• GCC 4.3.0 now enables auto-vectorization at the -O3 optimization level to better take advantage of x86 SIMD instructions. The auto-vectorization optimizations are further enhanced with the new vectorization cost model.

AMD has close relationships with operating systems and compiler teams at Red Hat and Novell, as well as with GNU tools developers.  AMD makes contributions to GCC and collaborates with these Linux software development ecosystem partners to ensure the availability of relevant software optimizations in the GNU toolset they distribute for 32-bit and 64-bit platforms including AMD platforms.  Optimized RPMs are available directly from Red Hat and Novell.

» Optimized GCC 4.3.2 RPMs for Fedora 10
» Optimized GCC RPMs for SLES 10, OpenSuSE 10.3, 11.0 and 11.1
» How to download, build, and install GCC from source available at the FSF on an Ubuntu 8.04 system

AMD provides links to web sites containing information, applications, software and other content that does not belong to or originate with AMD.  Said third party sites are not investigated, monitored or checked for accuracy, appropriateness, or completeness by AMD.  Further, AMD is not responsible for, and does not provide support for, any third party sites or any content posted on, available through or installed from the third party site.

AMD also directly distributes optimized RPMs for select versions of GCC.

GCC 4.2 Downloads
GCC 4.1.2 is an upgrade to the version of GCC shipped with RHEL5 and SLES10 SP1, and adds the following features:

• Support for SSE4a instructions included in AMD "Istanbul" CPUID Family 10h processors
• Support of Advanced Bit Manipulation instructions

» Download optimized rpms for GCC 4.2 now

GCC 4.1.2 Downloads
GCC 4.1.2 is an upgrade to the version of GCC shipped with RHEL5 and SLES10 SP1, and adds the following features:

• Support for SSE4a instructions included in AMD "Istanbul" CPUID Family 10h processors
• Support of Advanced Bit Manipulation instructions

» Download optimized rpms for GCC 4.1.2 now

» GCC Quick Reference Guide for optimized x86 code generation

» Compiler Usage Guidelines for AMD64 Platforms

» Software Optimization Guide for AMD Family 10h Processors

» AMD CodeAnalyst™ Performance Analyzer for Linux

» AMD Core Math Library (ACML) high-performance math library for 32-bit and 64-bit Linux with full implementation of BLAS, LAPACK, FFT, and other routines

» Framewave open source library of popular image and signal processing routines

» SSEPlus open source library of common SIMD operations that leverage current and proposed SSE

» More documentation about the AMD "Istanbul" Processor Family (CPUID Family 10h)