The evolution in the processor segment has been abysmal. If a few years ago we had large processors (in size), today in a SoC (System on Chip) we have the CPU and another set of sensors.

The neuromorphic CPU that offers the best performance for the world of Artificial Intelligence has recently been revealed. This small “brain” can perform 10 trillion operations per second (TeraOPs / s).

Optical neuromorphic CPU: More than 10 trillion operations per second

A new optical neuromorphic processor developed by Swinburne University of Technology is 1000 times faster than any previous processor. This processor, aimed at artificial intelligence (AI), can perform more than 10 trillion operations per second (TeraOPs / s).

The invention can revolutionize neural networks and neuromorphic processing in general. "This discovery was achieved with" optical micro-combs ", as was our world record for internet data speed revealed in May 2020," said Professor David Moss of Swinburne, one of those responsible for this project.

“Micro-combs” are new devices made up of hundreds of infrared lasers, all kept on a single chip. Compared to other optical sources, micro-combs are much smaller, lighter, faster and cheaper. The new innovation demonstrated by the Swinburne team uses a single processor, which simultaneously merges data in time, wavelength and spatial dimensions using a single micro-comb chip (small chip).

Cost and energy savings

According to Professor Arnan Mitchell of RMIT University, "technology is applicable to all forms of processing and communications" and will result in significant cost savings and energy consumption in the future.

At convolutional neural networks they were fundamental to the artificial intelligence revolution, but the existing silicon technology presents an increasingly bottleneck in processing speed and energy efficiency.

This advance shows how a new optical technology makes these networks faster and more efficient and is a profound demonstration of the benefits of interdisciplinary thinking.

Neuromorphic computing is characterized by the implementation of circuits that mimic the biology of the nervous system. Through this technology, the way computing works today, namely at the level of Artificial Intelligence, will become more evolved and adapted to the demands of the human being in its man-machine dynamics.

Read too…

Leave a Reply

Your email address will not be published.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