A GA based energy aware scheduler for DVFS enabled multicore systems

Multicore systems are prevalent now as high end computing systems for solving computationally complex problems. Energy consumed by these machines are enormous specially in instruction execution on the cores. It has been observed that if CPU cycle and latency cycle are properly managed, it is possible to save a good amount of energy. Dynamic voltage frequency scaling (DVFS) technique is often used to scale energy consumption at the cores. Job scheduling to the appropriate cores, in general, is an NP-hard problem. This work aims at effective use of DVFS technique at the instruction level and applies genetic algorithm, a popular meta-heuristics, for job scheduling at the appropriate core for optimal energy usage of multicore systems. Experimental results, on the benchmark data, exhibit that the proposed model is well scalable and energy efficient with acceptable performance tradeoff over other contemporary models.