Similarly,
Technological breakthrough ultra fast greener:
Artificial intelligence systems (AI) like GPT cat require a large amount of energy to operate. Moreover, To respond to this issue. Consequently, a team from the optical, photonic and lasers center (COPL) has designed an optical chip which allows you to quickly transfer a gigantic amount of data. In addition, As fine as a hair, this technology offers unequaled energy efficiency.
Published in the journalNature Photonicsthis innovation is based on the use of light to transmit information. Furthermore, Unlike traditional systems that only exploit the light intensity. For example, this chip also uses the phase of light, that is to say its gap.
By adding a new dimension to the signal, the system achieves unrivaled performance, while retaining a tiny size. However, “We go from a speed of 56 gigabits (GB) per second to 1000 GB technological breakthrough ultra fast greener per second. Therefore, ” said doctoral student Alireza Geravand, first author of the study.
– Technological breakthrough ultra fast greener
The research team sees a huge potential for learning AI models. For example, “With 1000 gigabits per second. you could transfer all training data-the equivalent of more than 100 million pounds-in less than 7 minutes, it’s time to prepare coffee,” he adds. Et tout cela ne consommerait que 4 joules, soit l’énergie nécessaire pour chauffer un millilitre d’eau d’un degré Celsius.
Innovation is based on ring microorsonators. These tiny silicon devices are able to handle light and encode information. The system is made up of two pairs of rings: one for intensity, the other for the phase.
Current AI data centers use dozens, even hundreds of thousands of processors, who communicate with each other like neurons in a brain. With a length of a few millimeters each. the necessary infrastructure quickly becomes enormous and the energy to supply it too. “We end up with a system that is kilometers long,” says the doctoral student. Thanks to their technology. the devices can communicate quickly and effectively, as if they were only a few meters from each other. This is a considerable advantage when AI needs keep growing.
This technology could be integrated into industry in the coming years. Companies like NVIDIA are already starting to use microorsonators, although they are limited to light intensity.
“10 years ago, our laboratory had demonstrated this technology. Today we take a new step. Perhaps in a few years. the industry will catch up with us and this technological breakthrough ultra fast greener innovation will be found on the market, ”concludes Alireza Geravand.
The other signatories of the study are Zibo Zheng, Farshid Shateri, Simon Levasseur, Leslie A. Rusch and Wei Shi.
[ Branchez-vous sur l’actualité de l’Université Laval! ]
Further reading: In Bosnia and Herzegovina, thirty years later, thousands of people celebrate the memory of victims – “Good evening Good evening!”: Jean-Philippe Wauthier has been hospitalized since Wednesday evening – Vaud: in prison for three years, he sees the charges collapse – Julian Alaphilippe, the bad news – Soon a “waste tax” for tourists?.
 like GPT…){kind=link}