GMA: a high speed metaheuristic algorithmic approach to hardware software partitioning for Low-cost SoCs

This paper presents a high speed and resource-efficient algorithm for the problem of Hardware Software Partitioning. The problem is modelled as a multi-dimensional global optimization problem with the aim of minimizing the area utilized, power dissipated, time of execution and system memory requirement of the implementation. An O(n) approach is proposed which significantly reduces the complexity of algorithms previously described in this domain, which can thus be implemented on resource constrained environments like low cost SoCs effectively. This enables speedy design decisions and henceforth, a faster development process. Then, detailed empirical analysis of the proposed algorithms is presented. A comparison of the execution time of the proposed algorithm with other major partitioning algorithms in the literature is also performed and the effectiveness of our approach is demonstrated. The time of execution is as low as 0.011 seconds, with deviation from optimal solution by less than 2%, for partitioning an algorithm consisting of 2500 blocks.