Integration of optical computation in processor architectures for automotive use cases

The continuous development of better processors has brought up significant improvements over the past years, but is nearing physical limitations and an increase in engineering efforts. With physical scale-effects becoming harder to realize and energy efficiency being a major need for industry and home-users, several ideas of next-generation computation are currently being researched to counteract the increase of computational demand. Those include, but are not limited to, neuromorphic, biological and optical computers, that might enable a new jump in computational offerings.

This thesis will take a deeper look at possibilities and background of optical computation and further analyze how this technology can be integrated into processor architectures. Whilst this integration is believed to enable more efficient computation of Artificial-Intelligence (AI) inference use cases, the author will explore technological opportunities and limitations of an integration into a RISC-V architecture. This analysis will provide an outlook on use cases, where optical accelerators will be beneficial and what automotive use cases it could impact.