Japanese researchers have developed a new type of light-based processor. Named as the Photonic Accelerator, or PAXEL, it seems to be set on becoming the corner milestone to make photonics finally on par with current electronic technologies.
Photonics is a field of technology that already features many important applications in telecommunications today. However, it has yet to break into the larger consumer tech industry due to its current level of technology still being a lot cruder to use in processors and other very tiny circuitry components.
Electronic circuits, on the other hand, are now facing a looming crisis over the maximum design limitations of current processors. Beyond the 5 nanometer range, is the theoretical maximum limit of what is known as Moore’s law. Basically, we can no longer shrink integrated circuitry further, because we are already reaching the level where quantum errors are bound to make binary values unstable.
PAXEL attempts to move above current limitations in photonics, as well as going beyond Moore’s law, by being designed with nanophotonics. Instead of measuring voltage levels to indicate a one (1) or zero (0) value in binary, it attempts to create a data measurement system using light intensity. Different light intensity data mapping can denote either one of the four major arithmetic operations, allowing for a completely alternative circuitry system that is both independent of electric energy, as well as being basically inapplicable to Moore’s law.
One very interesting feature that the researcher has introduced with PAXEL is what is called as “fog computing”. It is described as being similar to cloud computing, but is instead used very near within the originating light data pattern sources from “micro data-centers”. Hence fog, or “clouds on the ground”. Due to the local nature of this feature, the research also points out its potential to be integrated with upcoming 5G networks, in order to facilitate even better near-negligible latency PAXEL networks.
While this all sounds like a magical solution to all our computing problems. the PAXEL is not without its own share of challenges. Volatile data storage remains an issue for any theoretical photonic CPU. Your future nanophotonic processor may be able to compute with less latency than current high-end CPUs, but it is not capable of storing that light data pattern directly onto RAM without transition lag.
Another issue is that photonic processors require a complete overhaul of our current digital information processing and computational architectures. As you might have read earlier, photonics use light gradients to create data, unlike electronic processors that simply rely on voltage differences.
Building a fully integrated electronic-photonic processor, something that conceivably has to happen first, will be presented with challenges considering how the two architectures would mesh into one another.
Featured Image credit: Uwe Hermann via Flickr