thoughts on plasmonsters? would designing these to replace

>>60495079
Haven't heard of it before. Thanks anon.

>>60495079
Optical computing is on the leading edge of VLSI. Last I checked the tech isn't there yet, but it promises a lot of possible improvements

>>60495079
No. Assuming you have ref light, the wavelength is in the order of 800nm. Transistors are already much smaller than what you could physically accomplish with light.

Problem of miniaturization precision I'd take it.

>>60495282
You just need smaller light.

>>60495295
I know it's bait, but optics with UV is hard and you only gain a factor of 2 or so. And x-rays are hard to contain.

>>60495282
That's just visible light spectrum.

Extreme UV can go to 10nm, however I think these lights are bit too hard to control and probably early light computing devices that's fast as modern computing would have short life span, if shielding isn't done properly.

>>60495360
Actually, there are some approaches to overcome the wavelenght limit and reach nanoscale, by using noble metal particles. Light when hitting this particle will be contained around the surface of the particle, reducing its size to nansocale.

>>60495079
>>60495352
This. Very much this. Light is huge compared to transistors.

What we need is a way to interface computer chips together with light. We can get more bandwidth out and eventually there will come a point when we can't add any more pins.

Although optical computing is cool for other reasons. I'm too tired to autism about them right now

>>60495360
>>extreme UV
no. Just no. Extreme UV fucks shit up and it's hard as fuck to make an extreme UV light source.

>>60495508
>>Actually, there are some approaches to overcome the wavelenght limit
and that's what plasmonics does. Just like we can transmit large wavelength RF signal from coax, we can essentially make really tiny coax cables that operate at visible light frequencies.(No it's not actually coax)
https://en.wikipedia.org/wiki/Plasmonic_Circuitry

>>60495605
>>60495508
>>60495360
>>60495282
Why exactly does it have to be competitive with current transistor technology in terms of size?

The biggest reason why transistors have shrunk so much is to reduce power consumption. By making them smaller, less charge is required to turn on a data path. This means less thermal energy wasted in the CPU. I'm sure most of you understand the need for cooling in a CPU, most of you probably have aftermarket cooler in your tower.

Transistor speed is mostly limited by the thermal properties of silicon and copper at the moment, with smaller transistors only a way to dump less thermal energy.

So if a light transistor is 100 times the size of a typical transistor, that is only 100 times the area, or 10 times the length. That's already good enough for data centers. The only real thing stopping commercialization of this technology is how to manufacture it cheaply, and then how to interface with existing electrical systems.

>>60495079
Optic computers aren't appearing any time soon. In the short term we will replace silicon with germanium for example. Then maybe we will try alternative transistors, like vacuum transistors. Maybe then spintronics will take off. Optics though, I simply don't see any place for it, light simply doesn't interact with itself enough to be useful for computing.

>>60495765
Wrong, the finite speed of light is a major factor in bounding die size.

>>60495765
>The biggest reason why transistors have shrunk so much is to reduce power consumption
That's one of the reason.

The primary driver is to be able to pack more transistors onto the same die resulting in more powerful processors. The increased density allows for more complex processor design, moar coars or more parallel architecture.