A New Application Feature Analysis Approach for System-on-Chip Hardware/Software Partitioning

System-on-Chip is comprehensively applied in the image and signal processing field. With higher demands of customers and increasing complexity of System-on-Chip, System-on-Chip system level design issues become more critical. Hardware/software partitioning is an important step in the system level design. Application feature analysis and early design decision for hardware/software partitioning become a “must”. In this paper, we present a new application feature analysis approach. The approach excavates and utilizes application features including control flow, dataflow, computation, storage, and critical code features. And based on the analysis results, feature model is automatically generated to support hardware/software partitioning. The approach can excavate and utilize application features to improve partitioning algorithms and SoC performance. A case of JPEG codec indicates that this approach supports early design decision and helps to design space exploration in hardware/software partitioning.