AMD Optimizing C/C++ and Fortran Compilers (AOCC)

Overview

The AOCC compiler is designed for high-performance x86 CPU compilation of C, C++, and Fortran programming languages. It offers target-dependent and target-independent optimizations, with a particular focus on AMD "Zen" processors. These optimizations provide performance advantage for time-critical applications, especially in the field of high-performance computing (HPC). The AOCC compiler environment streamlines and speeds up the development and tuning process for x86 applications written in C, C++, and Fortran.

AOCC Features

• Built for applications targeting 32-bit and 64-bit Linux® platforms
• Leverages the robust LLVMTM Compiler Infrastructure
  • Based on LLVM 17.0.6 release (llvm.org, Nov 2023).
  • Optimized for AMD processors based on the AMD "Zen" core architecture, spanning AMD "Zen," AMD "Zen2", AMD "Zen3", AMD "Zen4", and AMD "Zen5" architecture generations.
  • Leverages Clang as the default front-end for C and C++ applications.
• Compliances
  • C17 is the default standard for C.
  • C++17 is the default standard for C++.
  • Debugging support for C, C++, and Fortran applications in compliance with DWARFv4 standard by default. DWARFv5 support is also available.
  • Offering Flang as the default Fortran front-end compiler supporting F2008 and Real128 features. Coarrays are not supported.
  • Supports OpenMP 5.0 for C and C++ applications, and OpenMP 4.5 for Fortran applications.
• Advanced Optimizations
  • Inter/Intraprocedural analysis and optimizations
  • SLP and loop vectorizations
  • Loop optimizations
• Debugging support for
  • C, C++, and Fortran applications in compliance with DWARFv4 standard by default. DWARFv5 support is also available.
  • OpenMP Debugging Interface (OMPD) in OpenMP host C and C++ applications.
  • CPU Offloaded codes in Fortran OpenMP.
• Supports Spack open-source utility for flexible package management, including use by AMD optimized application recipes.
• Other features:
  • -fPIC and -fPIE options made default since 4.1.
  • Default optimization level is -O2 since 4.1.

Notes

1. AOCC compiler binaries depend on Linux® systems having Glibc version 2.28 or later.
2. As -fPIC and -fPIE options are default since AOCC 4.1, location independent code is generated by default. With AOCC 4.1 or later, static libraries that are statically linked and not using -fPIC options so far must be rebuilt with AOCC with these default options. Else, it could display a linker error. Alternatively, use the options -fno-PIC/-fno-PIE with AOCC to disable generating location independent code.

What’s new in AOCC 5.0 (October 10, 2024)

• Based on LLVM 17.0.6 release (llvm.org, Nov 2023)
• Optimized support for AMD "Zen5" architecture
• Improved SLP and loop vectorization
• Improved LICM and loop optimizations
• Enhanced control/data flow optimizations
• Zen5 tuned AOCL-LibM 5.0 (AMD Math Library)

Download with End User License Agreement

For installation guidance, refer to README.

Resources and Technical Support

Documentation and Downloads

• AOCC User Guide
• AOCC Quick Reference Guide
• AOCC Release Notes
• Prior versions: AOCC Archive

Support

For support options, refer to Technical Support.

AMD Community

For moderated forums, refer to the AMD community.