A Multidimensional Program Node Modeling Approach Based on Multi-Level Analysis in Hardware and Software Partitioning

This paper proposes a multidimensional program node graph model based on multi-level analysis to address the issues of single granularity partitioning and low utilization of program structure information in the process of hardware/software (HW/SW) partitioning. By extracting the control and data transfer relationships between nodes at different levels, a multidimensional attribute graph model is constructed, which includes control flow information, function call information, and data transfer information. Experimental results demonstrate that the multidimensional attribute graph model satisfies the consistency requirements of the program and improves the flexibility HW/SW partitioning.