Superconducting CPU?

For the discussion of the sciences. Physics problems, chemistry equations, biology weirdness, it all goes here.

Moderators: gmalivuk, Moderators General, Prelates

morriswalters
Posts: 6904
Joined: Thu Jun 03, 2010 12:21 am UTC

Re: Superconducting CPU?

Postby morriswalters » Wed May 31, 2017 1:18 pm UTC

quantropy wrote:I don't think it would be that hard to prevent users from using the GPUs for their own purposes (well maybe not users who are really dedicated, but the effort they put in would probably be better spent earning money and buying the latest GPU board). I'm assuming that the a lot of the cost of GPUs is development cost, so manufacturers could afford to sell them cheaply if they were happy that this wouldn't affect sales of the latest version. Older GPUs tend to be less energy efficient, but this is being sold as a heater. Yes, it would cost quite a bit more than a simple fan heater, but you'd be surprised what people are prepared to pay for electric heaters with a fancy name: https://www.amazon.co.uk/d/ajl/Electror ... B00FEEU9US

I would see it as allowing the user to enter their WiFi details, but have no other user interaction.
Light a fire in your fireplace in the summer. Let me know how that works for you when you are running your AC. And Copper Bezels idea about perfect efficiency fails to account for the fact that an 800W computational load costs you somewhere between 60W and 200W additional load since we generate in AC and computers use DC. However if you feel you must, buy a bitcoin mining machine, and donate half the bitcoin to your favorite distributed computing charity to help them with their fixed costs. You get the heat and half the bitcoin to offset your losses. And the warm feeling you get thinking about your contributions to the betterment of society. You can pick it up a small one for about 500 USD's sans power supply.

User avatar
Copper Bezel
Posts: 2416
Joined: Wed Oct 12, 2011 6:35 am UTC
Location: Web exclusive!

Re: Superconducting CPU?

Postby Copper Bezel » Wed May 31, 2017 6:06 pm UTC

Isn't the energy lost from converting AC to DC the thing that makes power bricks and PSUs warm? Couldn't you just put the PSU inside the heater?

gmalivuk wrote:Yeah, what's the point of an IoT device that doesn't have gaping security holes?

If it isn't possible for someone to remotely burn your house down with this heater, it's trash.
So much depends upon a red wheel barrow (>= XXII) but it is not going to be installed.

she / her / her

quantropy
Posts: 191
Joined: Tue Apr 01, 2008 6:55 pm UTC
Contact:

Re: Superconducting CPU?

Postby quantropy » Wed May 31, 2017 6:37 pm UTC

HES wrote:You're proposing a heater where the user cannot change the heat. At least throw in some fancy smartphone temperature control to justify the markup.

Well, maybe it could be marketed as a super-duper way to control your heating, but as others have pointed out there are then security issues. Possibly it could be set to increase your house insurance just before it burns your house down.

User avatar
Eebster the Great
Posts: 2752
Joined: Mon Nov 10, 2008 12:58 am UTC

Re: Superconducting CPU?

Postby Eebster the Great » Wed May 31, 2017 6:55 pm UTC

It's certainly possible to make an inefficient heater. For instance, it could emit long wave radio or shine a light out the window. But it's pretty hard to accidentally make a wasteful heater.

And yes, energy lost during rectification is heat of course.

Frenetic Pony
Posts: 177
Joined: Fri Jan 27, 2012 6:31 am UTC

Re: Superconducting CPU?

Postby Frenetic Pony » Wed May 31, 2017 10:55 pm UTC

Eebster the Great wrote:It's certainly possible to make an inefficient heater. For instance, it could emit long wave radio or shine a light out the window. But it's pretty hard to accidentally make a wasteful heater.

And yes, energy lost during rectification is heat of course.


And part of the power output of any computer device today is actually high frequency sound! So all 800 watts is not only NOT used for computation, it doesn't even all come out as heat. Sometimes, depending on the component (and if it's going bad) the frequency gets low enough to hear. "Coil whine" as you might have seen it.

User avatar
Copper Bezel
Posts: 2416
Joined: Wed Oct 12, 2011 6:35 am UTC
Location: Web exclusive!

Re: Superconducting CPU?

Postby Copper Bezel » Thu Jun 01, 2017 12:24 am UTC

Vanishingly little, though. Like, it appears that 2-3 watts of sound is equivalent to a lawn mower engine.

Edit: And it still ends up as heat somewhere, of course, but probably not somewhere useful.
So much depends upon a red wheel barrow (>= XXII) but it is not going to be installed.

she / her / her

User avatar
Eebster the Great
Posts: 2752
Joined: Mon Nov 10, 2008 12:58 am UTC

Re: Superconducting CPU?

Postby Eebster the Great » Thu Jun 01, 2017 2:18 am UTC

Unless that sound is escaping to a meaningful extent through the walls of the room, it is still heating the room just as efficiently. That's why I clarified "light out the window," since any other light (as with sound) will simply heat the room anyway. Getting the energy outside of what you intend to heat is actually a challenge in itself. Making an inefficient heater is much more difficult than making an efficient one.

User avatar
Copper Bezel
Posts: 2416
Joined: Wed Oct 12, 2011 6:35 am UTC
Location: Web exclusive!

Re: Superconducting CPU?

Postby Copper Bezel » Thu Jun 01, 2017 2:25 am UTC

Yeah, that's a good point; I was thinking about the attenuation of the sound, then considered that it probably all had to do with the square of the distance thing and that the vibration really might "escape", and didn't think about the fact that sounds can be, you know, muffled.
So much depends upon a red wheel barrow (>= XXII) but it is not going to be installed.

she / her / her

morriswalters
Posts: 6904
Joined: Thu Jun 03, 2010 12:21 am UTC

Re: Superconducting CPU?

Postby morriswalters » Thu Jun 01, 2017 4:32 am UTC

Frenetic Pony wrote:And part of the power output of any computer device today is actually high frequency sound! So all 800 watts is not only NOT used for computation, it doesn't even all come out as heat. Sometimes, depending on the component (and if it's going bad) the frequency gets low enough to hear. "Coil whine" as you might have seen it.
Your making a distinction that doesn't exist. It a product of the action of the coil. And it's called inefficiency. But for any machine there is a minimum number of parts that it takes to make it operate. The heat is a product of inefficiencies throughout that minimal machine. And that value determines the power draw. Logic gates alone are insufficient to the purpose. There have to be coils.

The simple power equation for electricity is P=VI=I2R. Does anyone think that their computer, if it were drawing 1000 Watts, uses any different type of power or more or less power, than a 1000 Watt resistor designed to act as a heater? To reduce power draw you either reduce the voltage or reduce the resistance. Thus a superconducting computer. R near zero. Cut the voltage in half and reduce the power draw of your smartphone by a similar amount. And don't include rectification on board.
Eebster the Great wrote:Unless that sound is escaping to a meaningful extent through the walls of the room, it is still heating the room just as efficiently. That's why I clarified "light out the window," since any other light (as with sound) will simply heat the room anyway.
Which says something about the efficiency of the room, but not so much about the efficiency of your source. If you use a 95 percent efficient furnace, nobody will ask you if you leave your windows open. Good grief, and I promised that I would butt out. Sorry. It's a character flaw.

User avatar
Zamfir
I built a novelty castle, the irony was lost on some.
Posts: 7302
Joined: Wed Aug 27, 2008 2:43 pm UTC
Location: Nederland

Re: Superconducting CPU?

Postby Zamfir » Thu Jun 01, 2017 10:20 am UTC

Copper Bezel wrote:Vanishingly little, though. Like, it appears that 2-3 watts of sound is equivalent to a lawn mower engine.

Ah, that's 2-3 watts of electric speaker power, but only on the order of 0.01W of sound power... Regular speakers are not very efficient. 1 watt of sound power is more like a police car siren.

User avatar
gmalivuk
GNU Terry Pratchett
Posts: 25789
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

Re: Superconducting CPU?

Postby gmalivuk » Thu Jun 01, 2017 11:46 am UTC

Yeah if something is drawing 800W and is quieter than a machine gun, it's losing around 1% or less of that power to sound.
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

User avatar
Zamfir
I built a novelty castle, the irony was lost on some.
Posts: 7302
Joined: Wed Aug 27, 2008 2:43 pm UTC
Location: Nederland

Re: Superconducting CPU?

Postby Zamfir » Thu Jun 01, 2017 12:45 pm UTC

Another point of pedantry:

Making an inefficient heater is much more difficult than making an efficient one.

It's more that "heat_out/energy_in" is not a very good concept to evaluate the efficiency of heating. It makes a heat pump just as efficient as a resistor. Unless you restrict to electric energy, and then a comparison to non-electric systems becomes problematic. You need something like exergy analysis to account for different grades of energy.

And on the output side, it important to focus on the function of a heater, a step beyond simple production of heat. It's to keep something warm, or to provide a pleasant living climate, etc. Different heaters are not equally efficient at such a function, even if they generate the same kW output in heat.

There are fairly objective standards that describe thermal comfort. Temperature where the people actually are (in particular, not the ceiling). A good mixture of IR radiation and air convection. Avoid the presence of cold or hot drafts, or cold touchable surfaces and floors. or asymmetrical IR radiation (which makes the unlit side of your body feel cold, even if the air temperature is normal), etc.

User avatar
Eebster the Great
Posts: 2752
Joined: Mon Nov 10, 2008 12:58 am UTC

Re: Superconducting CPU?

Postby Eebster the Great » Fri Jun 02, 2017 3:02 am UTC

Yeah I'm using "efficient" in a very narrow sense that doesn't have much to do with how good a fit the heater actually is for a particular purpose.

User avatar
Copper Bezel
Posts: 2416
Joined: Wed Oct 12, 2011 6:35 am UTC
Location: Web exclusive!

Re: Superconducting CPU?

Postby Copper Bezel » Fri Jun 02, 2017 4:12 am UTC

Yeah, if we get into heater design, that's a whole other set of less interesting questions that apply equally to all types of heating elements. The question was in the (nonexistent) difference between a GPU and a big resistor coil all other things held equal. Though I suppose the GPUs may be bulkier and lead to relative inefficiencies in manufacturing cost.

Edit: Well, okay, the original question was the reverse of this whole scenario, but.
So much depends upon a red wheel barrow (>= XXII) but it is not going to be installed.

she / her / her

wumpus
Posts: 498
Joined: Thu Feb 21, 2008 12:16 am UTC

Re: Superconducting CPU?

Postby wumpus » Thu Jun 08, 2017 8:20 pm UTC

One thing I should point out is that I *think* a superconducting CPU would be amazingly faster (for some computations).

The point isn't power. The point is that your clockspeed is mostly determined by the RC (resistance time capacitance) delays in the circuit. If R drops to zero, all your gates (in a sufficiently small block [such as an ALU, but if it clocks fast enough you can make it pretty small) suddenly can zoom of to unbelievably fast (limited by power, instead of limited by leakage). You still would have big problems in that longer wires would be dominated by inductance, and I don't think losing resistance will help enough with inductance.

So a superconducting CPU would stand CPU design on its head, and we would see some really strange designs (think back to the 1970s/80s or so when designers suddenly could say "drop a PAL here). GPUs would really take off, and be even more "GPUy" in that memory latency would be mind-boggling and TFLOPS would be through the roof.

But I'd imagine the whole thing would use a lot more power. Basically because the power would be going into the switching (and it would probably require cooling breakthroughs or really weird design [think rotating the place where the calculation occurs] to keep the thing from melting the hot spots down.

Also 800W. Lets assume half that goes into the CPU. And assume that voltage is wildly high (1.5v). That means you have 267A (hopefully you can smooth out the clock events across the die, otherwise expect some sharp peaks) going in/out of that chip. It also has ~1000 pins (typically the four-digit number in the socket name stands for the number of pins), so you have maybe 200-250mA going through each pin. That doesn't sound so bad until you hear that each "pin" is so small that they can't really solder to it and need to use thermal paste... Somehow I don't think 200mA into such a tiny thing is a good idea.

Note: I think the real solution is to build the power supply on the chip - but that's really off the wall and I suspect will be put off until absolutely necessary.

User avatar
Soupspoon
You have done something you shouldn't. Or are about to.
Posts: 2484
Joined: Thu Jan 28, 2016 7:00 pm UTC
Location: 53-1

Re: Superconducting CPU?

Postby Soupspoon » Thu Jun 08, 2017 8:29 pm UTC

wumpus wrote:(typically the four-digit number in the socket name stands for the number of pins)

Chips used to be faaar simpler in the olden days: https://en.wikipedia.org/wiki/Socket_1

;)

Frenetic Pony
Posts: 177
Joined: Fri Jan 27, 2012 6:31 am UTC

Re: Superconducting CPU?

Postby Frenetic Pony » Thu Jun 15, 2017 8:47 am UTC

wumpus wrote:One thing I should point out is that I *think* a superconducting CPU would be amazingly faster (for some computations).

The point isn't power. The point is that your clockspeed is mostly determined by the RC (resistance time capacitance) delays in the circuit. If R drops to zero, all your gates (in a sufficiently small block [such as an ALU, but if it clocks fast enough you can make it pretty small) suddenly can zoom of to unbelievably fast (limited by power, instead of limited by leakage). You still would have big problems in that longer wires would be dominated by inductance, and I don't think losing resistance will help enough with inductance.

So a superconducting CPU would stand CPU design on its head, and we would see some really strange designs (think back to the 1970s/80s or so when designers suddenly could say "drop a PAL here). GPUs would really take off, and be even more "GPUy" in that memory latency would be mind-boggling and TFLOPS would be through the roof.

But I'd imagine the whole thing would use a lot more power. Basically because the power would be going into the switching (and it would probably require cooling breakthroughs or really weird design [think rotating the place where the calculation occurs] to keep the thing from melting the hot spots down.

Also 800W. Lets assume half that goes into the CPU. And assume that voltage is wildly high (1.5v). That means you have 267A (hopefully you can smooth out the clock events across the die, otherwise expect some sharp peaks) going in/out of that chip. It also has ~1000 pins (typically the four-digit number in the socket name stands for the number of pins), so you have maybe 200-250mA going through each pin. That doesn't sound so bad until you hear that each "pin" is so small that they can't really solder to it and need to use thermal paste... Somehow I don't think 200mA into such a tiny thing is a good idea.

Note: I think the real solution is to build the power supply on the chip - but that's really off the wall and I suspect will be put off until absolutely necessary.


You've got most of the right idea, though for heating to be an issue once you eliminate non useful work (a superconducting computer would literally only be doing "useful" work in the quantum mechanical sense) you'd have to pack wires, even tiny ones, incredibly close together. Probably a 3D design unlike today's 2D logic chips, and even then your efficiency and speed would be such that you'd have a modern a high end compute cluster sitting in your "wearable" gadget. Besides, modern server rooms already use some crazy cooling techniques, sea water pipes for room level liquid cooling and etc. Changing to higher efficiency isn't going to change the modern high end of cooling.

For anyone curious "superconducting computers" is actually a research topic. As are light based computers (all optical and etc.) Though frankly, from a practical standpoint they're a bit silly. Graphene is much further along, conducts close enough to the speed of light with high enough efficiency anyway that you'd not get much benefit going to a full optical computer beyond that, at least not for binary computing. Quantum computing is a different monster.

User avatar
gmalivuk
GNU Terry Pratchett
Posts: 25789
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

Re: Superconducting CPU?

Postby gmalivuk » Thu Jun 15, 2017 11:38 am UTC

(You know that the "et" in "et cetera" already means "and", right? Seems weird to see "and etc.")
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

User avatar
Copper Bezel
Posts: 2416
Joined: Wed Oct 12, 2011 6:35 am UTC
Location: Web exclusive!

Re: Superconducting CPU?

Postby Copper Bezel » Thu Jun 15, 2017 1:53 pm UTC

It's thanks to this site that I learned to write it as &c., and I will never stop doing this because it's ridiculously cute.
So much depends upon a red wheel barrow (>= XXII) but it is not going to be installed.

she / her / her

User avatar
Eebster the Great
Posts: 2752
Joined: Mon Nov 10, 2008 12:58 am UTC

Re: Superconducting CPU?

Postby Eebster the Great » Fri Jun 16, 2017 3:17 am UTC

Gotta go even more old school with ⁊c. Or modern with +c. Or symbolic-logical with ∧c. or ·c.

Frenetic Pony
Posts: 177
Joined: Fri Jan 27, 2012 6:31 am UTC

Re: Superconducting CPU?

Postby Frenetic Pony » Fri Jun 16, 2017 6:46 am UTC

gmalivuk wrote:(You know that the "et" in "et cetera" already means "and", right? Seems weird to see "and etc.")


Well shit, and etc.

It just doesn't seem grammatically correct in my head :oops:

User avatar
Eebster the Great
Posts: 2752
Joined: Mon Nov 10, 2008 12:58 am UTC

Re: Superconducting CPU?

Postby Eebster the Great » Fri Jun 16, 2017 10:27 am UTC

Frenetic Pony wrote:
gmalivuk wrote:(You know that the "et" in "et cetera" already means "and", right? Seems weird to see "and etc.")


Well shit, and etc.

It just doesn't seem grammatically correct in my head :oops:

Usually people use it that way when they think "et cetera" is one word.

quantropy
Posts: 191
Joined: Tue Apr 01, 2008 6:55 pm UTC
Contact:

Re: Superconducting CPU?

Postby quantropy » Fri Jun 16, 2017 10:46 am UTC

So and etc. is like having 'and and' in a sentence.

User avatar
Soupspoon
You have done something you shouldn't. Or are about to.
Posts: 2484
Joined: Thu Jan 28, 2016 7:00 pm UTC
Location: 53-1

Re: Superconducting CPU?

Postby Soupspoon » Fri Jun 16, 2017 11:27 am UTC

quantropy wrote:So and etc. is like having 'and and' in a sentence.
Yes, that is an example of and and and adjacent... And and and and wrongly so, in that case.

User avatar
Copper Bezel
Posts: 2416
Joined: Wed Oct 12, 2011 6:35 am UTC
Location: Web exclusive!

Re: Superconducting CPU?

Postby Copper Bezel » Fri Jun 16, 2017 3:04 pm UTC

Eebster the Great wrote:Usually people use it that way when they think "et cetera" is one word.

Usually spelled "eccetra". XD
So much depends upon a red wheel barrow (>= XXII) but it is not going to be installed.

she / her / her

User avatar
Eebster the Great
Posts: 2752
Joined: Mon Nov 10, 2008 12:58 am UTC

Re: Superconducting CPU?

Postby Eebster the Great » Fri Jun 16, 2017 9:25 pm UTC

Soupspoon wrote:
quantropy wrote:So and etc. is like having 'and and' in a sentence.
Yes, that is an example of and and and adjacent... And and and and wrongly so, in that case.

Wouldn't the sentence "I want to put a hyphen between the words Fish and And and And and Chips in my Fish-And-Chips sign" have been clearer if quotation marks had been placed before Fish, and between Fish and and, and and and And, and And and and, and and and And, and And and and, and and and Chips, as well as after Chips?

User avatar
Soupspoon
You have done something you shouldn't. Or are about to.
Posts: 2484
Joined: Thu Jan 28, 2016 7:00 pm UTC
Location: 53-1

Re: Superconducting CPU?

Postby Soupspoon » Fri Jun 16, 2017 10:02 pm UTC

Eebster the Great wrote:
Soupspoon wrote:
quantropy wrote:So and etc. is like having 'and and' in a sentence.
Yes, that is an example of and and and adjacent... And and and and wrongly so, in that case.

Wouldn't the sentence "I want to put a hyphen between the words Fish and And and And and Chips in my Fish-And-Chips sign" have been clearer if quotation marks had been placed before Fish, and between Fish and and, and and and And, and And and and, and and and And, and And and and, and and and Chips, as well as after Chips?

No. Because the preexisting '"'s force you to use "'"s. :P


Return to “Science”

Who is online

Users browsing this forum: No registered users and 6 guests