Fiber Optics

>>8661168
>personal PCs
You mean personal PC computers.

>>8661168

Electric fields move faster in copper than light in fiber optics.

>>8661289
lold. nice

>lets waste this electricity to generate light to move data 0.0000000000001 metres across a CPU

are you fucking retarded?

>>8661354

what about when they develop circuits that work with light instead of electricity

>>8661394
I actually designed a couple of logic components based on light intensity, they can be of an arbitrary base i.e. not binary but they would require a stupid sensitive method to measure light intensity, I'm not sure if the photoelectric effect would be enough.

>>8661168
intel made something similar few years ago, but was never released. it could achieve speeds of up to 50Gbps

>>8661289

Damn. The worst part is that I was trying to avoid that mistake.

>>8661311

Really? That's interesting. However, isn't the benefit of fiber optics the bandwidth?

>>8661354

I was more thinking in terms of throughput. So data transfer on the bus is not a limiting factor in a computer?

A follow up question. Do you know of any optical storage technologies? What prevents optical technologies from replacing tech that is known for being slow (e.g. harddrive)?

>>8661405

I see. That's very interesting. So, from what you are saying, it sounds like one huge hurdle is sensitivity and reliability.

I personally find it interesting that we have all of this LED and photovoltaic tech, but I have never heard of these technologies as having the potential to, even experimentally, replace traditional computer hardware.

So, I was just curious if there were some EE, CSE, or Physics specialists that had some insights into the limitations of using optical tech to replace traditional PC parts.

>>8661168
fiber optic cable is expensive to make

>>8661447
telecom anon here, commercial fibers can do 400Gbps max speed

in labs research groups have gotten Pb traffic

File: freakazoid_by_vergani.jpg (36KB, 600x676px) Image search: [Google]

36KB, 600x676px

>>8661447
>>8661483
>>8661566

Da-damn. The future sounds fast.

>>8661168
> https://en.wikipedia.org/wiki/Optical_computing
Probably similar in principle to what you're thinking of.

https://eandt.theiet.org/content/articles/2016/05/photonic-interconnects-computing-goes-on-a-light-diet/

File: optical interconnects.jpg (79KB, 367x326px) Image search: [Google]

79KB, 367x326px

>>8661168
That's called an optical interconnect. The problem is generating and sensing the light in a small form factor.

>>8661354
https://en.wikipedia.org/wiki/Interconnect_bottleneck
As we make transistors smaller, we need to pack wires going into the chip into a smaller space, this increases parasitic capacitance and signal delay.

Optics don't have this problem

>>8661394
light is big. about 40 transistors could fit in the wavelength of red light.
>>8661483
So are computer chips

Less power use and significantly lower error rate than silicon and copper which allows for M-Ary data transfer. Surely the question is when.

>>8661168

Optics are good (if not the best) for long distance transfer of information but I don't see the purpose of it in really close distances (such as PC architecture). You'd need amplifiers/converters at each entering/exiting node plus glass cables need to be really clean in order to function properly so cleaning your PC could be only done by professionals and let's not even mention the problem with dust that accumulates in PCs.

It's just not practical at the moment. Maybe some kind of revolution in technology will change that.

>>8661168
Mix some sodium chloride and put it into a thin hose. Put the hose in the freezer. Activate absolute-zero mode. Take the hose and run a current through it. Goals.

Check mate, physicists.

>>8661925
>light is big. about 40 transistors could fit in the wavelength of red light.
yet a single wave could carry more information than 40 transistors in one clock-cycle.
because there are a lot more hues to red than there are 1's and 0's in 40 transistors.

optical circuits WILL be the future of computation.

>>8662268
But you obtain a slower clock speed because the light has to travel larger distances between photonic elements.
Optical communication between chips makes sense, optical computing as the chips does not.

all this tech just for faster porn

>>8662268
>because there are a lot more hues to red than there are 1's and 0's in 40 transistors.
Are you retarded?

2^40 >> "hues of red" whatever that's supposed to mean

Even if we had sensitivity to a difference of 10^-6 nanometre in wavelength and used the entire visual spectrum 2^40 >> 400*10^6.

Surely you realise that there are engineering and possibly quantum mechanical limits to the continuous spectrum of light.