Networks-on-Chip: An Interconnect Fabric for Multiprocessor Systems-on-Chip

Interestingly, the request for scalable performance is being posed not only to high-performance microprocessors, which have been tackling this challenge for a long time, but also to embedded computing systems. As an example, systems designed for ambient intelligence are increasingly based on high-speed digital signal processing with computational loads ranging from  MOPS for lightweight audio processing,  GOPS for video processing,  GOPS for multilingual conversation interfaces, and up to  TOPS for synthetic video generation. is computational challenge has to be addressed at manageable power levels and aordable costs [BOE].