The Research Of Energy-Aware Scheduling Algorithm For Multi-core Processor

In recent years, the energy consumption of processor becomes increasingly prominent. From uniprocessor to multiprocessor, there is an increasingly of performance, however it consumes more energy, not only caused a heat dissipation problem which proves difficult to resolve, but also led to the falling of the reliability of the system. So, how to reduce the system energy consumption as much as possible while meet the time constraints is becoming an urgent problem of real-time energy-saving scheduling for multiprocessor.Recently, the researches focus on how to make full use of the slack time because of the actual execution time is far less than the worst execution time and the static slack time caused by the task completed before the deadline. They design speed scheduling algorithm to reasonably reduce the pressure/frequency of the processors to satisfy the time constraint and the energy-aware. This paper reduces the consumption caused by changing pressure/frequency of the processor when using slack time to reduce the speed.The main contents of this paper are organized as follows:(1) This paper proposes a new multiprocessor energy-efficient scheduling algorithm, which is based on the slack time reclamation and reduction of the energy consumption caused by switching frequency of the processor. First of all, use the radio of pre-executing tasks to get average task utilization rate to estimate the speed of the post-executing task. Then map the post-executing task to the processor which has the most similar to the estimate speed, to reduce the energy consumption of changing the speed. Then the current slack time is fairly shared by two tasks. It reduces the processor speed of the processor and minimizes the energy consumption. Through the theory and performance evaluation, compared to the algorithms which is proposed in the nearly three years, the new algorithms obtain better energy-saving effect.(2) Through the experiment, this paper analysis the influence factors, constraint condition and the applicability of the algorithm. Then proposes three improved algorithms TR-SS2(Load Balancing Strategy), ESR-SS2-CE and ESR-SS2-M, increasing the constraint condition of the speed, which sets the speed as the estimated speed and the maximum speed. Simulation results show that the improved energy-saving scheduling algorithms obtain a better effect in energy-saving.