Physics

A multilevel breakthrough in optical computing – faster, more efficient, and more robust memory cells

A concept image of the group’s photonic in-memory computing. Credit: Brian Long, Senior Artist, UCSB

For the first time, an international executive of electrical engineers has developed a new method for photonic in-memory computing that could make optical computing a reality in the near future.

The team includes researchers from the University of Pittsburgh’s Swanson School of Engineering, the University of California, Santa Barbara, the University of Cagliari, and the Tokyo Institute of Technology (now Tokyo University of Science). Their work was published today in the journal Nature Photonics under the title “Ultra-rugged integrated nonreciprocal magneto-optics for photonic in-memory computing.”

The research was co-coordinated by Nathan Youngblood, an assistant professor of electrical and computer engineering at Pitt, and Paulo Pintus, formerly at the University of California, Santa Barbara and currently an assistant professor at the University of Cagliari in Italy. It’s research. Associate Professor Yuya Shoji of Tokyo University of Science.

Until now, researchers have been limited in developing photonic memories for AI processing, gaining important properties such as speed while sacrificing others, such as energy usage. In this article, an international team presents a unique solution that addresses the current limitations of optical memory, which has not yet been able to combine non-volatility, multi-bit storage, high switching speeds, low switching energy, and high endurance in a single platform. We have demonstrated our solution.

“The materials used to develop these cells have been available for decades. However, they are primarily used for static optical applications such as on-chip isolators, rather than platforms for high-performance photonic memories. It has been used for many years,” Youngblood explained.

“This discovery is a key technology that will enable faster, more efficient, and more scalable optical computing architectures that can be programmed directly in CMOS (complementary metal oxide semiconductor) circuits, thus enabling today’s computer technology. This means that it can be integrated into

“Additionally, our technology demonstrated three orders of magnitude better durability than other non-volatile approaches at 2.4 billion switching cycles and nanosecond speeds.”

The authors propose a resonance-based photonic architecture that utilizes nonreciprocal phase shifts in magneto-optic materials to implement photonic in-memory computing.

A common approach to photonic processing is to multiply a rapidly changing optical input vector by a matrix of fixed optical weights. However, encoding these weights on-chip using traditional methods and materials has proven difficult.

By using magneto-optical memory cells composed of cerium-substituted yttrium iron garnet (Ce:YIG) nonuniformly integrated on a silicon microring resonator, the cells can be moved like sprinters running in opposite directions on a track. allows light to propagate in both directions.

Control and calculate the speed of light

“It’s like the wind is blowing towards one sprinter and helping the other sprinter run faster,” said Pintas, who led the experimental work at the University of California, Santa Barbara. Explained.

“By applying a magnetic field to the memory cell, we can control the speed of light differently depending on whether the light flows clockwise or counterclockwise around the ring resonator. This allows us to , providing an additional level of control not possible with traditional non-magnetic materials.” ”

The team is currently working on scaling up from single memory cells to larger memory arrays that can support even more data in computing applications. They point out in their paper that irreversible magneto-optic memory cells provide an efficient non-volatile storage solution that can provide unlimited read/write endurance with sub-nanosecond programming speeds.

“We also believe that future advances in this technology may allow us to exploit various effects to improve switching efficiency,” added Tokyo’s Shoji. “Also, new manufacturing techniques and more precise deposition using materials other than Ce:YIG may further increase the potential of non-magnetic materials.” – Mutual Optical Computing. ”

More information: Paolo Pintus et al., Ultra-rugged integrated non-reciprocal magneto-optics for photonic in-memory computing, Nature Photonics (2024). DOI: 10.1038/s41566-024-01549-1

Provided by University of Pittsburgh

Citation: Multilevel Breakthroughs in Optical Computing – Faster, More Efficient, and Robust Memory Cells (October 23, 2024), https://phys.org/news/2024-10-multi- breakthrough- Retrieved October 23, 2024 from opticalfastefficiency.html

This document is subject to copyright. No part may be reproduced without written permission, except in fair dealing for personal study or research purposes. Content is provided for informational purposes only.

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button