would a high temperature superconducting CPU be the next

File: 1476368636152.jpg (199KB, 1280x720px) Image search: [Google]

199KB, 1280x720px

dumb maki poster

>>57155372
One word, my friend:
>carbon nanotubes

>>57155394
but carbon nanotubes can't perform superconducting. if anything we should look at yttrium nanotubes.

File: 1459117808761.png (699KB, 622x570px) Image search: [Google]

699KB, 622x570px

>>57155381
Do not bully Akarin!

>>57155398
I honestly don't know anything about them. I just always see them brought up when it comes to semiconductor stuff.

*smooches akari-chan's adorable loli cheeks*

>>57155401
Who? I'm only seeing air.

Most of the losses are from transistor switching, not resistance.

I have some gallium nitride mosfets already, switching losses are around 10x less than silicon and they withstand higher temperatures.

As far as I know the only reason it's not used is cost, it's still cheaper to try and squeeze more out of silicon than to switch.

>>57155401
Could you please fuck off back to /a/?

>>57155372
4th season when?
>high temperature
>superconducting
keks

File: 1472853729156.png (59KB, 1300x736px) Image search: [Google]

59KB, 1300x736px

>>57155631
Dont bully akarin or akarin posters

>>57157788
Kill yourself

>>57155394
>carbonnanotubes

what did he mean by this?

how is Akarin so cute

File: rope.jpg (256KB, 1000x1502px) Image search: [Google]

256KB, 1000x1502px

>>57157794
Leave now! This is a technology board on an anime webpage! Need I remind you of the history of /g/?

File: tumblr_inline_nf8uk5cxQX1t43dcu.jpg (23KB, 304x395px) Image search: [Google]

23KB, 304x395px

if you're referring to like very high temp very high clock speed computing i wouldn't count on it. push a current cpu to ~5ghz and the electrons are already just barely making their way across the entire die at the speed of light inbetween the ticks, any higher than around 7-8ghz is physically impossible for that very reason.

File: cewig.jpg (97KB, 854x480px) Image search: [Google]

97KB, 854x480px

>>57157802
Planned obsolescence. The wet dream of Intel.

File: 1396952465511.jpg (83KB, 637x476px) Image search: [Google]

83KB, 637x476px

>>57157856
The electrons hardly move at all. The waves in the electrons are what move fast, and those only get around at less that a third of the speed of light.
I would imagine the issue is temperature and stability, not the speed of light.

>>57155372
Are slightly dopey faces that are still super cute the next step in what anime can do?

File: 1449788758944.png (289KB, 824x542px) Image search: [Google]

289KB, 824x542px

>>57158018
By stability I mean that the wavelength of the electric field is too narrow / small to be properly handled by the clock skew. With higher clock speeds comes exponentially more power draw, so overheating and consumption becomes an issue as well.

>>57155631
Could you please fuck off back to re ddit?

>>57158018
>I would imagine the issue is temperature and stability, not the speed of light.
Not him but signal propagation is a real problem with all our current chips. Even if we had magic transistors that produced no heat, it still wouldn't get us much further.

>>57158128
The propagation velocity of the electric fields only push 30% of light speed. And that's the speed of the wave, not the electrons.
So you are completely right but he was mostly wrong.
And for now, we still have a ways to go until we get to the smallest possible architecture size, so we ought to let that run its course before moving on to other means of speeding up computing, in my opinion.
Beyond that, the solution will come in reducing resistance and increasing the simplicity of the chips' layouts.

File: 51978658_p0.jpg (94KB, 345x396px) Image search: [Google]

94KB, 345x396px

>>57158128
Also we can push the signals up to frequencies like 8GHz+ without propagation being an issue, at that point it's limited to what the clock skew can manage. Frequencies that high are pretty easily achievable (with pretty high draws ofc) but the wavelength is too narrow for the clock skew to keep up with, I think.
I'm only self educated in electronics, so I may very well be incorrect, but this is my thinking.

>>57158294
>Also we can push the signals up to frequencies like 8GHz+
In a single bulldozer module, with the rest of the chip turned off. Or an ancient celeron. Either way, we're not talking about a complete, modern chip.

>>57158345
That doesn't change the fact that the propagation speed isn't the limiting factor. As I said, we are limited now by architectural size. It is very difficult for any current clock skew to keep up with a frequency as high as 8GHz, because the wavelength is so narrow and the architecture is not designed to handle this.
I'm not arguing that 8GHz is achievable on modern hardware, it isn't. Im arguing that propagation speed isn't yet a limiting factor and that electrons don't go the speed of light inside your PC

>>57158400
>That doesn't change the fact that the propagation speed isn't the limiting factor
Erm, yes it does. If you limit the amount of space the clock has to propagate through, then you eliminate clock propagation as a problem. I thought this was pretty self explanatory.
>and that electrons don't go the speed of light inside your PC
That's not being disputed, feel free to drop that whenever it suits you.

>>57158441
Dropped, and okay I think I understand a little better. As I said, I'm not educated on this, so I'm probably wrong.
It would seem, then, that shrinking manufacturing process and simplifying layout is the only solution to this issue, assuming we don't drastically change the way CPUs work?