Measuring Efficiency and Executability of Allocation and Scheduling in Multi-Processor Systems-on-Chip - Abstract


Multi-Processor Systems-on-Chips (MPSoCs) are becoming increasingly complex, and mapping and scheduling of multi-task applications on computational units is key to meeting performance constraints and power budgets. Abstract models of system components and deployment of advanced algorithmic techniques for the optimization problem can provide for fast design space exploration and for optimal solutions. In [3] we have proposed an efficient hybrid approach for solving the problem based on an integrated Constraint Programming (CP) and Integer Programming (IP) solution where IP performs the allocation and CP the scheduling. The hybrid solution exploits problem decomposition and chooses for each subproblem the best algorithm for solving it given its structure. We have proved the efficiency of the hybrid approach w.r.t. both stand alone CP and IP approaches. In this paper, we go a step further and exploit an accurate MPSoC virtual platform for capturing mismatches between problem formulation and real-life systems, and for assessing their impact on expected performance. We therefore evaluate the efficiency and the executability of the solution found by the algorithm using a MPSoCs platform simulator.





Quick Links
 Last updated Mar. 21, 2004