The design of embedded systems is often subject to strict requirements concerning various aspects, including real-time performance, power consumption and die area. For mobile devices especially, power consumption is often the most important issue. In order to meet these requirements an adequate system architecture needs to be designed and the embedded software needs to be optimized. For complex applications, this optimization can be fairly difficult and requires an in-depth analysis to reveal the most energy demanding parts of the software. In this paper, we describe a method and its implementation for cycle-accurate software analysis on the function-level, which is based on a RISC processor power model. This analysis method has been used during the implementation of a video coding application in order to show its applicability.