Optical computing

[bryanl]

Will optical computing ever offer advantages at the chip level or even board level, as opposed to the interconnect level?

 

[intx13]

Will optical computing ever offer advantages at the chip level or even board level, as opposed to the interconnect level?

Optics aren't used for computing at the interconnect level, but for data transfer. Optical transducers take space and (for now) the rest of the a computer system uses copper, so I don't see optical paths replacing copper traces for data transfer on a motherboard any time soon.

As for computation, quantum computing typically uses optics as the quantum channel. Quantum computing may one day appear as an add-on to a traditional PC (for things like offloading FFTs or other cycle-finding problems into hardware) but I don't think it will really replace the semiconductor-based binary system we know and love.

 

[Revolution 11]

One reason optics may not be adapted is not just cost or difficulty of manufacturing but also the lack of scalability. The current semiconductor economic model revolves around regular upgrades. If you create circuitry that uses light, can you really scale this up without increasing cost/area/power? What would be the method of improvement?