Fast Multi-Dimension Multi-Choice Knapsack Heuristic for MP-SoC Run-Time Management

Since the application complexity is growing and applications can be dynamically activated, the major challenge for heterogeneous multi-processor platforms is to select at run time an energy-efficient mapping of these applications. Taking into account that many different possible implementations per application can be available, and that the selection must meet the application deadlines under the available platform resources, this optimization problem can be modeled as a Multi-dimension Multi-choice Knapsack Problem (MMKP), being NP-hard. Not only algorithms for exact solution, but also state-of-the-art heuristics for real-time systems, are still too slow for run-time management of multi-procesor platforms. This paper provides a new greedy heuristic for finding near-optimal solutions of the MMKP, being fast enough for the considered environment. The main contribution of this heuristic is: (1) the derivation of the Pareto sets from the initial MMKP to reduce the search space, (2) the sorting of all Pareto points together in a single two-dimension search space, where (3) a fast greedy algorithm solves the MMKP. Experiments show that our heuristic finds solutions close to the ones obtained by the fastest state-of-the-art heuristics (within 0% to 0.4% of the solution value), in just a fraction of the execution time (more than 97.5% gain on a StrongARM processor) and can run in less than lms for multi-processor problem sizes

I. Introduction

Since the application complexity is growing and applications can be dynamically activated, the major challenge for multiprocessor platforms is to select at run time an energy-efficient mapping of these applications. To alleviate the run-time decision making, and to avoid conservative worst-case assumptions, a two-phase customized run-time management concept can be used.