基于嵌入式软件算法级能耗问题研究

Based on the embedded software algorithm level energy consumption problem research

能耗已经成为嵌入式系统设计中一个重要的约束条件。嵌入式系统是典型的软件驱动执行系统,硬件的电路活动直接导致系统参数功耗,而软件中的指令执行和数据存取等操作底层的微处理、总线、Cache、存储器和I/O接口等硬件的活动都会间接的导致系统参数功耗。在现代"低碳经济"的背景下嵌入系统的功耗已经引起人们关注的一个重点。而软件设计早期对高层所作的功率耗评估和优化对整个系统的的能耗影响最为显著。本文通过构建算法级能耗估算模型,并通过实例采用神经网络算法、遗传算法等进行能耗求解,同时在求解过程中进行能耗分析。

energy consumption has become an important part of the embedded system design constraints.Embedded system is a typical software driver execution system, the hardware circuit of the activity as a direct result of the power consumption of the system parameters, and the instruction execution in the software and data access operations such as at the bottom of the microprocessor, bus, Cache, memory and I/O interface hardware activities will indirectly lead to the system parameters such as power consumption.In the modern ＂low-carbon economy＂ under the background of the power consumption of embedded system has been one of the focuses of attention.And software design of the early on top of the power consumption of evaluation and optimization is the most significant influence on the energy of the whole system.In this paper, by constructing level algorithm to estimate the energy consumption model, and by using the neural network algorithm and genetic algorithm for solving the energy consumption, and analysis of energy consumption in the solving process.