An Efficient Traffic-Based Routing Algorithm for 3D Networks-on-Chip

摘要

Network-on-Chip (NoC) is proposed to solve the communication challenges in multiprocessor System-on-Chip architecture. 3D NoC uses through silicon vias to stack multiple 2D silicon layers and obtain better performance. However, 3D NoC has to design an efficiently adaptive routing algorithm to overcome huge traffic load problems. An adaptive routing algorithm is composed of routing function (RF) and selection function (SF). RF decides a set of deadlock-free candidate paths, and SF chooses a proper routing path from those candidates. In this paper, we propose a novel traffic-based routing algorithm, TBRA, to achieve better performance. TBRA combines some advantages from different routing algorithms and selects adaptive routing algorithms according to the traffic load. Furthermore, it also collects two hops data of networks to determine the better routing path. Simulation results show that the proposed scheme significantly improves the transmission latency, the throughput, and the energy consumption.