| dc.description.abstract | Energy efficiency in computing is becoming more and more important. With the rise of smart phones, a whole new industry was born where having a more energy efficient system would mean longer battery life and an edge over the competition. It has also recently been an area of interest in High-Perfomance Computing (HPC). This has fuelled research and development of heterogeneous multi-core architectures, utilizing different CPU cores to do different tasks.
Utilizing heterogeneous architectures fully is a challenge both for the hardware and software engineers. Online judging systems are platforms where users can compete and learn while solving problems, getting feedback on correctness and efficiency of their submissions. Climbing Mont Blanc is an online judging system focusing on energy efficiency on heterogeneous multi-cores, and is to our knowledge it is the only such system measuring energy efficiency, aiming to provide an environment for education and practice in energy efficient programming.
The CMB system currently reports time, energy and energy delay product (EDP) per submission. To assist users in performance tuning their solutions, and to give a better picture of what the program execution looked liked, some more detailed low-level statistics were wanted as user feedback. In addition, some general system architectural improvements were needed to improve stability and ease of development. This thesis focuses improving the system with regards to these goals. | |
