Pre-mapping Algorithm for Heterogeneous MPSoCs

Abstract

A Multiprocessor System-on-Chip (MPSoC) composed of different types of processors is known as heterogeneous MPSoC. This kind of MPSoC based on Network-on-Chip (NoC) is a promising target architecture to fulfill requirements of high processing and communicating rate, enabling the execution of several tasks at the same time. Among the challenges in current heterogeneous MPSoC design is the partitioning of application tasks aiming for the energy saving and for a fair load balance. This work's contribution is twofold: (i) a partitioning algorithm is developed, and (ii) the evaluation of using partitioning as a pre-mapping task is explored. This work analyzes and compares stochastic and new heuristic partitioning algorithms for obtaining low energy consumption and for load balance when applied to tasks partitioning onto heterogeneous MPSoC. In addition, simulated results indicate that the static partitioning technique could be applied to application tasks before mapping activities in order to improve the quality of the static or dynamic mapping design and also to minimize the processing time.