多核系统中基于温度限制的节能调度算法研究-计算机应用技术专业论文.docxVIP

华 华 中 科 技 大 学 硕 士 学 位 论 文 II II Abstract In recent years, real-time system has been widely used in the fields such as aviation, military, transportation and industry for its fast response and high reliability features. However, with the continuing development of manufacturing technology of VLSI, power density and energy consumption augment seriously on the processors, which lead to increased heat dissipation and decreased reliability of systems. Therefore, the problem of saving energy consumption of modern processors and constraining temperature becomes major concerns in the computer science area. Currently, there exits many solutions to the energy savings and temperature-aware problem, one of the most promising methods is to achieve energy efficiency and low temperature through improved scheduling algorithm. Dynamic voltage scaling has been proven to be an effective energy-saving technique, which can make the processors work on an appropriate voltage by the adjustment of supply voltage, the temperature problems can be solved through restricting the supply voltage before scheduling. This study targets at the scheduling algorithm of the dependency-constrained real-time tasks in the environments of multiprocessor system on chip. A temperature-constraint static energy-efficient scheduling algorithm is proposed by integrating a static energy saving scheduling algorithm and a temperature optimization scheduling algorithm. our algorithm takes advantage of the dynamic voltage scaling technology to reduce the energy consumption and the peak temperature of the whole system. The results from the experiments show that the proposed scheduling algorithm, taking into account the energy consumption and temperature factors, can save more energy than the temperature optimization scheduling algorithm. When compared with the energy optimization algorithms, the proposed algorithm is able to effectively minimize the peak operating temperature and increase the reliability of the system