Scientists have provided an exciting new perspective on the next steps in the future development of fast, energy-efficient computing systems that use light instead of electricity to process and store information – introduction of hardware stimulated directly by human brain activity.
A team of scientists, including Professor C. David Wright from the University of Exeter, has examined the potential for future computer systems using photonics instead. conventional electronics.
The article is published today (January 29, 2021) in the prestigious journal Photonics nature.
The study focuses on potential solutions to one of the world’s most important computing problems – how to develop computer technologies to process this data in a fast and energy-efficient manner.
Contemporary computers are based on von Neumann’s architecture in which the high-speed Media Processing Unit (CPU) is physically separated from the much slower program and data memory.
This means that computing speed is limited and power is lost with the need to move data to and from memory and processor over limited bandwidth and energy efficient electrical interconnections – known as botail von Neumann.
As a result, it is estimated that more than 50% of the power of today’s computer systems is consumed directly in this movement around data.
Professor C David Wright, from the University of Exeter’s Department of Engineering, and one of the study’s co-authors, explains “Clearly, a new approach is needed – one that could be linked main functions of computer information processing and memory, one that can directly incorporate in hardware the ability to learn, adapt and evolve, and one that gets on removing high-speed electrical interconnections and reducing speed. ”
Photonic neuromorphic computing is one such approach. Here, signals are communicated and processed by the use of light rather than electricity, allowing for higher bandwidths (processor distances) and significantly reducing energy loss.
In addition, the researchers will try to make the computer hardware itself isomorphic with a biological processing system (brains), by developing tools to directly report the basic functions of brain neurons. and synapses, then connect them together in networks that can offer fast, parallel, variable processing. for artificial intelligence and machine learning applications.
The latest in ‘brain-like’ photonic computing, and its potential future development, is the focus of an article entitled “Photonics for artificial intelligence and neuromorphic computing” published in the prestigious journal Photonics nature with a leading international team of researchers from the US, Germany and the UK.
###
Photonics for artificial intelligence and neuromorphic computing, BJ Shastri et al., Photonics nature, doi: 10.1038 / s41566-020-00754-y
Disclaimer: AAAS and EurekAlert! they are not responsible for the accuracy of press releases posted to EurekAlert! by sending institutions or for using any information through the EurekAlert system.