Latest Blog Post - Current Limitations and Potential Extensions of MAGEEC
This post briefly discusses some of the main contributions and limitations of the MAGEEC project and suggests several directions for future work.
MAGEEC
MAGEEC aims to combine work on compilation options which save energy with work on machine learning to create a compiler framework, capable of generating code that has improved energy efficiency. MAGEEC is an open project with active community involvement, and to participate, use the project wiki, in depth updates on the project can be found on the project blog, with ongoing discussion on the mailing list.
The project has four key goals:
- It will optimize for energy
- It will use physical measurement of energy usage, not models.
- It will be compiler agnostic—initially targeting GCC and LLVM
- It will deliver a working system, not just research papers.
MAGEEC is supported by the Technology Strategy Board of the UK government under its Energy Efficient Computing Initiative. It is a joint project between the open source compiler company, Embecosm and Bristol University’s Department of Computer Science.
The project started in June 2013 and ran for 18 months. The work continues under the TSERO project, which aims to apply the same techniques to save energy in High Performance Computing (HPC) systems. This website is preserved as a record of the research on the MAGEEC project.
