RANC: A Residue Arithmetic Nanophotonic Computer
Abstract
Due to the end of Moore's law and Dennard scaling, feature reduction and higher speed of clocking are seizing to be the source for higher computer performance.Therefore, it is of paramount interest to explore alternative technologies and architectures for the post-Moore's law era of computing. This project aims to build an integrated photonics computing system (from device to architectures) based on the residue number system (RNS) to achieve orders of magnitude improvements in computational speed per watt over the current state-of-the-art. The objectives of this research project are to: Offer potential transformative insights by exploring new materials for strong enhancements of light-matter-interactions. Explore attojoule per bit efficient and GHz-fast optical switching devices. Design and demonstrate compact 22 switches that are basic building blocks for optical residue arithmetic functions. Enable a novel approach to the design and evaluation of an entire class of optical compute engines based on residue arithmetic leading to multi-purpose computing. Explore co-design principles that relate device technology to the switch, the network architecture and the routing algorithm and methodology. Emulate and evaluate performance and accuracy using well-accepted community benchmarks. Pave the way for rapid and agile prototyping by enabling insights for advanced manufacturing on a silicon photonics platform. Enable the collective synergistic experience of the PI's, who are well established in their fields, to explore innovative nanophotonic computing paradigms.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 30, 2023
- Accession Number
- AD1230875
Entities
People
- Tarek El-Ghazawi
Organizations
- George Washington University