Finding optimal hardware/software partitions

Abstract

Most previous approaches to hardware/software partitioning considered heuristic solutions. In contrast, this paper presents an exact algorithm for the problem based on branch-and-bound. Several techniques are investigated to speed up the algorithm, including bounds based on linear programming, a custom inference engine to make the most out of the inferred information, advanced necessary conditions for partial solutions, and different heuristics to obtain high-quality initial solutions. It is demonstrated with empirical measurements that the resulting algorithm can solve highly complex partitioning problems in reasonable time. Moreover, it is about ten times faster than a previous exact algorithm based on integer linear programming. The presented methods can also be useful in other related optimization problems.

Multi-core model checking and maximum satisfiability applied to hardware-software partitioning

“…Here, we refer to the relevant work of Mann et al (2007) in optimization, who modified a branchand-bound algorithm to speed up the execution time. Sapienza et al (2013) used multiple criteria decision analysis to solve the HW-SW partitioning problem.…”

“…The mathematical modeling of the HW-SW partitioning problem was taken from Arato et al (2003) and Mann et al (2007).…”

“…All ILP and GA parameters were presented and discussed in and . In particular, the ILP and GA algorithms were chosen since they were also used in another related work, as Arato et al (2003) and Mann et al (2007), thus facilitating the comparison to the present techniques (that we propose in this paper). Figure 5 shows ESBMC pseudocode with the same constraints and conditions placed on ILP and GA. Two values must be controlled to obtain the results and to perform the optimization.…”

“…The setup is described in Section 5.1, while Section 5.2 describes all benchmarks that were used for performing the experimental evaluation. Section 5.3 reports a comparison among MATLAB, ESBMC, ESBMC-SS, ESBMC-PS, ESBMC-PB, and ESBMC-νZ using a set of standard HW-SW partitioning benchmarks, as presented by Mann et al (2007).…”

“…Since the first decade of 2000s, two main paths have been tracked to solve the HW-SW partitioning problem: to find the exact solution of a optimization problem, as presented by Mann et al (2007); and to use heuristics to speed up performance time, as presented by Arato et al (2003) and Arato et al (2005). It is worth mentioning that using heuristics the final solution is not necessarily an optimal global solution.…”