对于传统的平面结构,三维片上网络(3D NoC)具有更好的集成度和性能,在单芯片内部可以集成更多的处理器核。3D NoC作为2D NoC的结构拓展,在性能提高和低功耗设计方面更具优越性,成为多核系统芯片结构的主流架构。映射就是应用某种算法寻找一种最优方案,将通信任务图的子任务分配到NoC的资源节点上,保证NoC的通信能耗最小。参照2D NoC的研究方法,提出了针对3D网格NoC的通信能耗模型,采用蚁群算法实现了面向通信能耗的NoC映射。实验结果表明,面向不同网络规模的3D网格NoC平台,蚁群映射同随机映射相比,通信能耗降低可以达23%~42%。
Comparing with the traditional planar structure, 3D NoC has the better performance and integrated density. It can be integrated with large numbers of cores in a single chip. As 2D NoC structure extension, 3D NoC can improve performance and reduce energy. It is becoming a main solution for big multi-core system chips. Mapping means to search for an optimal solution applying some algorithms, and assigns the tasks of the communication task graph to the NoC resource nodes to ensure the minimum of NoC communication power consumption. The energy-aware model of 3D Mesh NoC was presented based on the research of the 2D NoC, and the ant colony algorithm to achieve NoC mapping. Experiment results show that, comparing with the random mapping, the communication power consumption of ant mapping with different sizes of the 3D Mesh NoC platform, has decreased considerably about 23 %- 42 %.