Scaling Photonic Tensor Cores with Unary and Homodyne Designs

TL;DR

This paper analyzes five photonic tensor core designs, highlighting how circuit ordering, unary encoding, and homodyne accumulation influence scalability and parallelism in photonic computing.

Contribution

It provides a comparative analysis of five designs and identifies key techniques that enhance scalability and parallelism in photonic tensor cores.

Findings

Unary encoding and homodyne accumulation significantly improve scalability.

Circuit ordering impacts the efficiency of photonic tensor cores.

Homodyne accumulation offers the strongest path to higher parallelism.

Abstract

We analyze five photonic microring tensor core designs with a common optical power model. The results show that circuit ordering, unary encoding, and homodyne accumulation shape scalability, with the last two offering the strongest path to higher parallelism.