Proceedings Article | 10 September 2019
KEYWORDS: Computing systems, Computer architecture, Nanotechnology, Nanophotonics, Nanoelectronics, Nanoelectromechanical systems, Telecommunications, Spintronics, Plasmonics, Computer simulations
Future computing requires new science and technology for processing, memory, and interconnections. Nanoscale devices built out of new materials are the most promising option to overcome this challenge. Nanophotonic, nanoelectronic, nanomagnetic, and nanoelectromechanical technologies, such as spintronics, plasmonics, and nanoelectromechanical systems (NEMS), are some of the technologies currently in the nascent stages of development that hold strong promise to enable computing structures technology to provide building blocks for energy-efficient computing and data systems of the future. Investigating the underlying nanoscience, overcoming the
engineering hurdles to convert these technologies in order to build integrated devices, and designing new
computing architectures harnessing such nanotechnologies are the primary scientific goals of this
paper.
We seek a comprehensive approach, based on innovative nanoscience and technology studies, to conduct a clean-slate redesign of computing architecture and system design, simulation, and prototyping.
The impact of such studies can be immense.
1. Energy-efficient and scalable computing: The nanophotonics, nanoelectronics, nanomagnetics and nanoelectromechanical systems show strong promises as building blocks with extremely low-power consumption, and will help create energy-efficient, scalable, and miniature computing systems.
2. Balanced and optimized performance: While Amdahl’s law suggests that a system with balanced computation, memory and communications perform the best under most circumstances, today’s computing systems are typically misbalanced by more than two orders of magnitude. Massively parallel interconnections by nanophotonics and nanoelectronics bring the computing systems much needed communication bandwidth for more balanced and optimized operation for energy efficiency.
3. Fundamental changes in computing architecture: The new architecture with massively parallel interconnects enables a ‘flat’ computing architecture where locality and hierarchy considerations are no longer necessary. The emerging nanoscale science and technology offer a clean-slate approach that fundamentally changes the computing architectural design through parallel optical nanowaveguide and electrical nanowire interconnects.
4. Programmability: The new ‘flat’ computing platform greatly facilitates parallel programming and invites scientists to solve grand challenge problems.
