Migration from electronics to photonics in multicore processor

Abstract

Twenty - first opportunities for Gigascale Integration will be governed in part by a hierarchy of physical limits on interconnect. Microprocessor performance is now limited by the poor delay and bandwidth performance of the on - chip global wiring layer. This thesis is envisioned as a critical showstopper of electronic industry in the near future. The physical reason behind the interconnect bottleneck is the resistive nature of metals. The introduction of copper in place of aluminum has temporarily improved the interconnect performance, but a more disruptive solution will be required in order to keep the current pace of progress, optical interconnect is an intriguing alternative to metallic wires. Many - core microprocessors will push performance per chip from the 10 gigaflop to the 10 teraflop range in the coming decade. Pin limitations, the energy cost of electrical signaling, and the non - scalability of chip - length global wires are significant bandwidth impediments. Silicon nanophotonic based many core architecture are introduced in order to meet the bandwidth requirements at acceptable power levels.