XPS: EXPL: CCA: Optical Data Containers

University Of Rochester, Rochester NY

Investigators

Abstract

Computer systems today are all built around what is called the von Neumann architecture where the computation subsystem and the memory subsystem are separately designed and considered. This model has served the industry well for several decades. It does, however, create a bottleneck (referred as the von Neumann bottleneck) between the two subsystems. The increasing scale and dependence on data in modern computers put more and more stress on the von Neumann bottleneck, limiting computer speed and increasing energy costs. New technologies to address this problem can significantly improve all computers including modern data centers and supercomputers that are shouldering ever more responsibilities in science, medicine, engineering, business and commerce, and everyday life of an average citizen. This project is an effort to develop one such technology. The research centers around a novel nano-scale device recently developed by one of the investigators of the project. The device is called a coupled-disk resonator or CDR for short. CDR contains two flat disks very closely placed together. The fascinating property is that, thanks to technology, these disks can be shrunk down to nano-scale and placed in close proximity precisely. As a result, they can interact with light in a variety of controllable ways that allow computer designers to build superior optical communication systems than possible today. More excitingly, the CDRs can be manipulated to perform (simple) computations directly without having to convert optical signals first to electronic signals as is the case today. The potential outcome is not only high-performance communication channels, but smart ones that can significantly mitigate the von Neumann bottleneck. In this project, the investigators will study and experiment with the designs of both the device and the system built with these devices in an effort to bring this potential technology closer to reality.