International Journal of Computer Science Issues

More than a traditional journal...

The Network-on-Chip (NoC) Architectures have been proposed as a revolutionary methodology to the communication of multi-core embedded processor in a chip. Interconnection topology plays an important role in communication locality of future CMPs. We suggest extracting the locality communication of sub tasks of applications. Indeed it is essential to provide a suitable NoC topology that exploits the locality of communication resource fully. Due to the tradeoff between the performance and area constraint in on-chip network designs, we have developed in this paper a transactional level model NoC router of hybrid Mesh-Star-Ring. The new hybrid topology called Hybrid Locally Mesh Globally Star-Ring (HLMGSR) use the Star-Ring and mesh graphs as the basic building blocks. The experimental results show that for a network size equal to 128, the proposed hierarchical topology HLMGSR provides an area increase of 7% over the mesh. For a network size which is greater than or equal to 64 nodes, the lowest average latency is provided by the proposed hybrid topology for different injection rates. For a network size greater or equal to 64 nodes with the uniform traffic pattern with a lower injection rate, the results show that the HLMGSR provides the highest throughput.