基于ARM处理器的嵌入式软件能耗统计模型

ARM-Based Embedded Software Statistical Energy Model

提出了一种嵌入式软件能耗的统计模型,包括处理器、存储器和I/O控制器等硬件单元产生的能耗,然后通过分析ARM指令周期数的规律,设计了指令周期数的相应计算方法,该方法能够快速地计算软件运行时处理器产生的能耗。在高精度指令级嵌入式软件能耗模拟器HMSim中进行了模型实现。实验结果表明,该模型的能耗计算结果与实际仪器测量结果的误差在10%以内,可较准确地反映软件实现方式对系统能耗的影响程度。

The energy consumption of embedded software has become a key factor in embedded system design, and it is a fundamental work of analysis and optimization of energy consumption to measure the energy consumption of embedded sol,ware. This paper proposes a statistical model of embedded software energy consumption, including the energy consumptions of processor, memory and I/O controller, etc. Then, by analyzing the instruction cycle of ARM instruction set, a method for computing the instruction cycle number is designed so as to rapidly calculate the processor energy consumption of embedded software. This model has implemented in a high-precision instruction-level energy simulator HMSim. The experimental results show that the error rate of embedded software energy consumption estimated by this model is less than 10% compared with that measured by an electronic instrument. The achieved results also reflect how the embedded system energy consumption is influenced by different software designs.