The software is expected to automatically utilise newly developed programmes for embedded processors in mobile devices.
IBM has partnered with the IBM Haifa Research Lab, Israel; the University of Edinburgh and ARC International, UK; CAPS Enterprise and INRIA, France, for the project.
Mike O’Boyle, professor of computer science at the University of Edinburgh and project coordinator for Milepost, said: The Milepost solution uses artificial intelligence and machine learning to understand what kind of compiler optimisations are optimal for use with each new hardware design. This will help completely automate compiler construction and enable more rapid code design of hardware and software, dramatically reducing time to market in these systems.
The project consortium has released a prototype version of its software at the GNU Compiler Collection (GCC) Summit showcasing preliminary results. IBM claims that the project could improve the performance of a complier by 10% within one month.
Bilha Mendelson, manager of code optimisation technologies at the IBM Haifa Research Lab, said: Milepost is realising the vision of customised hardware with tailor fit software. Aside from shorter design cycles, Milepost opens new opportunities by enabling engineers to leap ahead and work with more experimental hardware. Opening the compiler infrastructure and combining it with machine learning techniques enables us to generate several sets of optimisations sequences for each hardware application area.
The Milepost GCC version is expected to be released in June 2009 and will be available in the open source community. At the end of a three-year programme, Milepost partners expect to release a new version of their compiler optimisation software into the GCC main product.
