Version 5.0 of Open64 compiler improves performance
The developers of the Open64 compilers have released version 5.0 of the tool, with improved performance, bug fixes and changes to the infrastructure of the compilation system. Open64 is an open source optimising compiler for x86-64, IA-32 and IA-64 platforms. Historically, Open64 is derived from SGI's Pro64 compiler for MIPS architectures; versions of the compiler for MIPS and other architectures such as CUDA and PowerPC are available from other sources. The main release of Open64 concentrates on Intel and AMD architectures and offers pre-built C, C++ and Fortran 95 compilers.
The developers say that the new release has a range of optimisations enabled, with better structure optimisation and improved vectorisation. The code generation engine has improved support for SSE 4.x, AVX and FMA instructions with "comprehensive support and tuning" for the AMD Bulldozer processor added. The developers have also taken the opportunity to deprecate support for the gcc 3.x frontend and IRIX, which will not be supported in the next release. The quality of the release has also been improved with 95 bugs, some long-standing, being fixed. Full details of the changes are available in the release notes.
The developers say that they will continue to update the frontend support for GCC/G++ with the latest versions and want to focus on improving performance, especially on multicore processors. They also expect to be able to provide static analysis and security checking tools, along with an optimisation framework.
Open64 compilers are available as binaries in rpm or tar archives and as GPL-licensed source code from the project's download page.