Automating Hardware/Software partitioning using dependency Graph

Hardware/Software (HW/SW) partitioning is a major step in embedded system design flow. This paper studies a new method of automated HW/SW partitioning. The main goal of this method is to improve performance in terms of the execution time and consumed area. Proposed algorithms are useful for partitioning using a dependency graph. This Graph models complete dependencies and flow of a design. We applied our method to a high-level description of MP3 decoder as a case study. The results show that a performance-improved automatic partitioning can be achieved using this method.