Experimental photonic processor to be assembled in Russia

< br />
MOSCOW, August 9The assembly of an experimental model of a Russian photonic processor, which processes information hundreds of times faster than modern digital neural networks based on traditional semiconductor computers, will begin in August and will be completed by the end of this year, said project participant, professor of the Department of Technical Cybernetics at Samara University named after Korolev, Doctor of Physical and Mathematical Sciences Roman Skidanov.
«The implementation of our project is going according to plan, the creation of an experimental sample of the photonic processor is now in the stage of case assembly. At the moment, all the main elements of the experimental sample have been manufactured, and this month, in August, we will begin assembly,» Skidanov said.

«A decision was made to use a different laser in the experimental sample — a diode type, it is more compact and has less coherence, which should improve the processor's characteristics. Future tests and experiments will show how much better. The assembly is planned to be completed and tests conducted by the end of 2024,» he added.

A demonstration sample of the processor was created by specialists from the Samara University named after Korolev within the framework of the scientific program of the National Center of Physics and Mathematics (NCPM), implemented with the support of the state corporation Rosatom. The processor operates on the basis of a new, photonic component base, in which information is transmitted by particles of light (photons), and not by electrons, as in conventional computers.

A sample of the photonic processor was created in the interests of implementing a «megascience» class photonic computing machine at the NCFM by 2030. According to the project, the machine's performance will be record-breaking and will reach 10 to the 21st power of operations per second. Such a «megascience» installation will make it possible to solve applied problems in processing large amounts of data and obtain fundamental results in the field of artificial intelligence and machine learning. A specialized processor already today makes it possible to recognize huge amounts of data in large video streams.
The analog photonic computing system allows for the analysis and recognition of objects hundreds of times faster than modern digital neural networks based on traditional semiconductor computers. This is especially important for the operational analysis of so-called hyperspectral data, which initially represent significant amounts of information. It was previously reported that the recognition reliability during the first experiments on a demonstration sample was 93.75%. The accuracy and reliability of recognition in the experimental sample should increase due to the selection of components with improved characteristics. It was also reported that the prototype of the installation is planned to be ready in 2025.