Research On Low-power Technique Based On Heterogeneous Multi-core Processor Mobile Equipment

With the development of communication technology,applications of Mobile Internet become more and more widely.The number of mobile equipment especially smart phone users is also growing rapidly.In order to meet the users' demands for performance,heterogeneous multi-core system comes into being and becomes a mainstream of mobile equipment system.While meeting the performance requirements,it also brings power consumption problem with the increment of applications,which not only reduces the effective working time of the equipment,but also increases the heat dissipation,affecting the user experience seriously,so low-power design has become one of the main tasks of heterogeneous multi-core system design.For heterogeneous multi-core systems with large amount of real-time tasks,the low-power scheduling algorithm is an effective way to reduce the power consumption of the system and becomes a hot issue.The system model must be established before the study of low-power scheduling algorithm.On the basis of system model,the low power scheduling algorithm based on heterogeneous multi-core system often adopts partition method to assign tasks to each processing unit,which transforms the multi-core scheduling problem into single-core scheduling problem and then performs single-core low-power task scheduling.To research the low-power scheduling algorithm of heterogeneous multi-core system,this thesis has done the following work:Firstly,establishes a relatively perfect system model.This model takes into account the heterogeneous multi-core processor,multi-device and multi-task coexistence situation,and reduces the power consumption by the combination of dynamic voltage and frequency scaling(DVFS)technology and dynamic power management(DPM)technology.Secondly,studies the application of genetic algorithm in partitioning method.The classical Max-min and Min-min partitioning algorithms can easily lead to reduce the real-time performance of the system when the number of tasks is large or the task execution time is long,while the partition algorithm based on genetic algorithm can further reduce the utilization of the processing unit and improve the real-time performance of the system in both cases.By simulation it can be seen that the real-time performance of the partitioning method based on genetic algorithm is higher than that of Max-min and Min-min under the same conditions.Finally,for the single-core low-power scheduling algorithm,studies the inter-frame scheduling algorithm.The traditional intra-frame scheduling algorithm doesn't consider the idle time relationship between frames,so the power consumption cannot minimize.Inter-frame scheduling algorithm is divided into single-task scheduling and multi-task scheduling.The single task scheduling algorithm first uses the DPM strategy to determine the task begin execution time,and then uses the DVFS strategy to determine the task execution frequency,thus achieving the lowest global energy consumption;The multi-task scheduling algorithm first uses the DVFS strategy to determine the task execution frequency,when a frame is scheduled,the DPM policy is used to determine the start time of the task,thus making the idle time energy of the device optimal.The simulation results show that the inter-frame task scheduling algorithm can further reduce the power consumption of the system compared with the intra-frame task scheduling algorithm.