A multinational team of researchers has developed a photonic processor that uses light instead of electronics and could help usher in a new dawn in computing.
Current computing relies on electrical current passed through circuitry on ever-smaller chips, but in recent years this technology has been bumping up against its physical limits.
To facilitate the next generation of computation-hungry technology such as artificial intelligence and autonomous vehicles, researchers have been searching for new methods to process and store data that circumvent those limits, and photonic processors are the obvious candidate.
Featuring scientists from the Universities of Oxford, Münster, Exeter, Pittsburgh, École Polytechnique Fédérale (EPFL) and IBM Research Europe, the team developed a new approach and processor architecture.
The photonic prototype essentially combines processing and data storage functionalities onto a single chip – so-called in-memory processing, but using light.
“Light-based processors for speeding up tasks in the field of machine learning enable complex mathematical tasks to be processed at high speeds and throughputs,” said Münster University’s Wolfram Pernice, one of the professors who led the research.
“This is much faster than conventional chips which rely on electronic data transfer, such as graphic cards or specialised hardware like TPU’s [Tensor Processing Unit].”
Led by Pernice, the team combined integrated photonic devices with phase-change materials (PCMs) to deliver super-fast, energy-efficient matrix-vector (MV) multiplications. MV multiplications underpin much of modern computing – from AI to machine learning and neural network processing – and the imperative to carry out such calculations at ever-increasing speeds, but with lower energy consumption, is driving the development of a whole new class of processor chips, so-called tensor processing units (TPUs).
The team developed a new type of photonic TPU capable of carrying out multiple MV multiplications simultaneously and in parallel. This was facilitated by using a chip-based frequency comb as a light source, which enabled the team to use multiple wavelengths of light to do parallel calculations since light has the property of having different colours that do not interfere.
“Our study is the first to apply frequency combs in the field of artificially neural networks,” said Tobias Kippenberg, Professor at EPFL
“The frequency comb provides a variety of optical wavelengths which are processed independently of one another in the same photonic chip.”
Described in Nature, the photonic processor is part of a new wave of light-based computing that could fundamentally reshape the digital world and prompt major advances in a range of areas, from AI and neural networks to medical diagnosis.
“Our results could have a wide range of applications,” said Prof Harish Bhaskaran from the University of Oxford.
“A photonic TPU could quickly and efficiently process huge data sets used for medical diagnoses, such as those from CT, MRI and PET scanners.”