eduzhai > Applied Sciences > Engineering >

A Design Methodology for Post-Moores Law Accelerators The Case of a Photonic Neuromorphic Processor

  • Save

... pages left unread,continue reading

Document pages: 4 pages

Abstract: Over the past decade alternative technologies have gained momentum asconventional digital electronics continue to approach their limitations, due tothe end of Moore s Law and Dennard Scaling. At the same time, we are facing newapplication challenges such as those due to the enormous increase in data. Theattention, has therefore, shifted from homogeneous computing to specializedheterogeneous solutions. As an example, brain-inspired computing has re-emergedas a viable solution for many applications. Such new processors, however, havewidened the abstraction gamut from device level to applications. Therefore,efficient abstractions that can provide vertical design-flow tools for suchtechnologies became critical. Photonics in general, and neuromorphic photonicsin particular, are among the promising alternatives to electronics. While thearsenal of device level toolbox for photonics, and high-level neural networkplatforms are rapidly expanding, there has not been much work to bridge thisgap. Here, we present a design methodology to mitigate this problem byextending high-level hardware-agnostic neural network design tools withfunctional and performance models of photonic components. In this paper wedetail this tool and methodology by using design examples and associatedresults. We show that adopting this approach enables designers to efficientlynavigate the design space and devise hardware-aware systems with alternativetechnologies.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...