Arbitrated IP Core Based Dynamic Task Mapping Algorithm for Networks-on-Chip

摘要

Mapping of the tasks onto the core is the main aim of the paper. So there is need to design such an algorithm that the tasks should not wait in the queue for long time. Our main focus is to design an algorithm that should concentrate on the load balancing, such that the problem of overheating of the cores is reduced. For this purpose different researchers have given different mapping algorithms. It would be the best solution if the designed algorithm could balance the load dynamically over the grid of cores. In this paper, we have mapped the tasks onto the cores such that performance of the network improves along with balanced load distribution. In experimental results, the proposed AICM approach is compared with sequential mapping considering 5X5 mesh topology.