Skip to Main Content
Networks-on-chip (NoCs) address the challenge to provide scalable communication bandwidth to tiled architectures in a power-efficient fashion. The 2-D mesh is currently the most popular regular topology used for on-chip networks in tile-based architectures, because it perfectly matches the 2-D silicon surface and is easy to implement. However, a number of limitations have been proved in the open literature, especially for long distance traffic. Two relevant variants of 2-D meshes are explored in this paper: high-dimensional and concentrated topologies. The novelty of our exploration framework includes the use of fast and accurate transaction level simulation to provide constraints to the physical synthesis flow, which is integrated with standard industrial toolchains for accurate physical implementation. Interestingly, this work illustrates how effectively the compared topologies can handle synchronization-intensive traffic patterns and accounts for chip I/O interfaces.