The Dynamic FM Technology Research Base On Linux System

Dynamic FM technology is a hot area of research in computer power and it is also an important part of the computer modules. Under the premise of guaranteed processing capacity, dynamic FM technology effectively reduces the power consumption of the system. In this thesis, we study the Linux operating system dynamic FM technology in order to reduce power consumption of the computer system.In this thesis,we discuss the background and significance of the Linux operating system dynamic FM technology and study the present situation at home and abroad. There are many problems in today’s methods of the CPU power obtain, such as computational complexity, complicated operation and inaccurate calculation. Based on the Linux operating system power management mechanism analysis, we design and implement a new CPU power acquisition software to solve the problems of CPU power to obtain. This software enables to obtain a period of time CPU power consumption by using the power consumption calculation model. Compared to the other methods of CPU power obtain, this software can not only be able to accurately obtain information on any period of time CPU power consumption but also operate easily. At the same time, it provides data support for the low-power study.In the Linux operating system, the CPU frequency is the only parameter that affects the load of the system. By the way, the CPU frequency is closely related with the system’s power consumption. Factors that affect the Linux operating system load have two aspects, which are the CPU utilization and the system number of executable processes. In the ondemand Dynamic FM policy, it needs a short time to reach up to the stable system load. In order to reduce the power consumption of Linux operating system, we study and analyze the common prediction algorithms. We choose binary linear regression equation as the prediction algorithm and apply it to the CPUFreq module of the Linux operating system. By acquiring the system CPU utilization and the number of system executable processes, this method can predict the processor load information timely and accurately, compared to the ondemand Dynamic FM policy and the AIBD prediction algorithm which is presented by Kang Qing. At the same time, it can reduce processor power consumption by setting the processor frequency accurately.The significance of Binary linear regression prediction algorithm dynamic FM strategy is that it can predict the next time the processor load information accurately and timely and set the processor operating frequency accurately to achieve the purpose of reducing power consumption. This method ensures the processor performance while reducing unnecessary power waste caused by the frequency setting.