Embedded system partitioning with flexible granularity by using a variant of tabu search

Various techniques to partition a system into hardware and software blocks have been proposed in the past. Most of these techniques use some form of control flow graphs (CFG) and employ optimization algorithms like simulated annealing or tabu search to reach an optimal solution. A partitioning method presented in this paper partitions a CFG representation by employing a variant of tabu search, which uses a dynamic tabu list. Fixed tabu list has been employed by most of the conventional algorithms. Our method works with a flexible level of granularity and it merges CFG nodes into partitioning objects under a defined set of rules. An initial partitioning object is selected and improved on subsequent iterations to find the best solution that satisfies the given constraints. The performance of the proposed method is compared with simulated annealing and conventional tabu search-based approaches that shows promising results.