Optical computing is being explored in several areas, including:
- Optical neural networks:
Analog in nature, artificial neural networks are a natural fit for
optical platforms, which inherently excel at executing the intensive
computations required for tasks like image recognition and natural
language processing. Various startups are deploying a range of methods
(discussed further below) to build neural networks and AI processors.
These include Lightmatter
and Lightelligence in the silicon photonics domain, Cognifiber in
multi-core fiber systems and Lumai in free-space optics (FSO).
- HPC applications involving a high number of iterations:
Computations like combinatorial optimization and simulations require a
large number of iterations to achieve exact results. Here, the inherent
parallelism of light helps to reduce compute time and power consumption.
Solutions in this space include: LightSolver’s all-optical FSO
processor, Microsoft’s hybrid electro-optical FSO machine and NTT’s
Coherent Ising Machine (CIM), which leverages fiber optics.
- Quantum computing: In the quest for a practical
qubit, photons have emerged as strong candidates due to their resistance
to environmental noise and ability to operate at room temperature.
Companies like PsyQuantum, Xanadu and QuantumSource are actively
developing photon-based quantum computers. These systems leverage the
unique properties of photons to enable scalable and efficient quantum
processing, though challenges like loss, weak interactions and
scalability remain critical hurdles.
https://www.eetimes.com/the-evolution-of-optical-computing-part-2/
Optinen tietojenkäsittely voi tietyissä sovelluksissa tarjota reippaasti tehoa edullisesti.
