Green Power Aware Approaches For Scheduling Independent Tasks On A Multi-Core Machine

The energy consumption of large Information and Communications Technology structures such as data and computation centers along with the corresponding carbon footprint are on the rise. Green computing has become an indispensable solution to face the resulting economical and environmental challenges. Powering these centers with renewable energy sources is however a challenge since these sources cannot guarantee a constant power supply due to their fluctuating power production. We here tackle the problem of scheduling independent tasks on a multi-core machine within a predicted renewable power envelope that varies over time. We evaluate the complexity of different instances of the problem from a theoretical point of view. We propose several heuristics, including genetic algorithms, and we conduct experiments to assess their performance. For some particular cases we compare the performance of these heuristics to optimal solutions.