Power modeling of an embedded RISC core for function-accurate energy profiling

The design of embedded hardware/software systems is often underlying strict requirements concerning various aspects including real time performance, power consumption and die area. Especially for mobile devices power consumption is an important issue. In order to meet the requirements and design a well-adapted system, the software parts need to be optimized and an adequate hardware architecture needs to be designed. Software optimization can be rather difficult for complex applications especially if targeting multiple aspects simultaneously. It is described how an existing profiling tool for fast and accurate performance and memory access analysis has been extended for energy analysis The development and integration of the required power model of a processor is described as well as an exemplary analysis of a software implementation of an H 264 video decoder. This paper is structured as follows: Chapter II gives an overview of existing power models and tools. The architecture of the MEMTRACE profiling tool is described in Chapter III. Chapter IV discusses the development of a measurement-based power model and its integration in the profiling tool. The energy profiling has been used for analyzing the software implementation of an H 264 video decoder and the results are depicted in Chapter V. Finally, Chapter VI summarizes the results of this work and gives some future prospects