Generalized Execution Time Estimation

Abstract

While developing embedded devices a significant challenge is to choose the appropriate computing architecture. Especially during early design stages providing measurable metric on the performance demands to implement a specified algorithm is required. Usually this includes a large amount of target dependency, like the chosen micro-controller platform. This work aims on providing a generalized method to estimate the processing time of a dedicated algorithm, when being run on a variety of hardware choices. The focus is on a fast exploration of the hardware design space in order to provide guidance for selecting a suitable processing platform.