The Future of Computing: The Renaissance of Analog and Digital Specialized Processors

Abstract

Since the advent of Moore's law, the evolution of computing has followed a straight line made of general-purpos electronic digital processors. In this era, the benefits of any specialised hardware were doomed to be overtaken by the next generation of general-purpose hardware. Now, we are on the verge of a change: while Moore’s law is reaching its limit, there is a pressing demand for novel computing, communications, and information processing capabilities. 
For the first time in the digital age, if we want to keep pushing the limits in communication and compute, we must reshape the architecture of our systems.

This seminar traces the history of computing from the "Golden Age" of general-purpose hardware to the modern renaissance of specialized analog and digital solutions. We will examine the limitations of the "bigger is better" AI paradigm and explore how analog electronic and photonic circuits are being rediscovered as high- efficiency alternatives. Next, the talk will present tasks that are enabled by specialised hardware and software solutions. These include: (I) the use of optical processors for space applications; (II) quantum applications of photonic circuits; and (III) artificial intelligence in computer network edges through hardware-aware design.

About the Speaker

Lorenzo De Marinis received the B.S. and M.S. degrees in Electronic Engineering from the University of Pisa (2017, 2019) and a Ph.D. from Scuola Superiore Sant’Anna (SSSA) in 2022. He was a Visiting Scholar at the Aristotle University of Thessaloniki (WinPhoS Lab) in 2022, researching machine learning for optical links.
Currently, he is an Assistant Professor at SSSA under the National Quantum Science and Technology Institute. His work focuses on the design of photonic integrated circuits for quantum, neuromorphic, and space applications. His broader research interests encompass analog computing, photonic-electronic circuit co-design, and the application of in-network AI for cybersecurity.