Mccaula718

Distinguished
Sep 11, 2006
34
0
18,530
Is it possible for a cpu to run too cold? I keep seeing all these things about cpu's being cooled by hydrogen, helium, dry ice, etc. How is a cpu able to work at -1xx*? I always compared a cpu to a car engine, it runs better when its cooler, but haven't you tried to start an engine in the winter, doesn't work as well. Someone please explain or direct me to another topic or editorial.
 
There is a point where you can cool the CPU so much that it would actually slow down the electrons and in so doing, kill performance, but I'll go out on a limb and say that neither you nor I have access to that kind of cooling technology. This would require getting it down to absolute zero which is 0K (-273.15 °C or -459.67 °F) something they try to achieved in high tech labs.
Scientists have made great advancements in achieving temperatures ever closer to absolute zero (where matter exhibits odd quantum effects). In 1994, NIST achieved a record cold temperature of 700 nK (billionths of a kelvin). In 2003, researchers at MIT eclipsed this with a new record of 450 pK (0.45 nK).
But for regular purposes, the cooler the better. :D
Hope this helps you,
Ninja
 

JTLJudoMan

Distinguished
Aug 7, 2006
8
0
18,510
There are the various cold bugs and so forth that prevent chips from working when you go to super cold temperatures.

I am not sure what the current ratios of silicon to germanium or whatever they're using to dope with these days... But at some point the resistance will probably go to 0... At this point the processor would cease to function. Transisters need the resistance to properly function. Without it they wouldn't work anymore... so yeah it doesn't even matter if you slow down the electrons because the physical properties of the transisters fails long before the effect of "slowing" electrons can be taken into account. so yeah... its more that the properties that drive transisters dont apply anymore.
 

bluntside

Distinguished
Mar 22, 2006
744
0
19,010
isnt silicone plastic type in a way? Wouldnt that contrapt the core making it brittle and shatter internaly? My hypothesis is that if your cpu gets too hot it will crash, same concept if it gets too cold. How about the NANOCOOLER?
 

chuckshissle

Splendid
Feb 2, 2006
4,579
0
22,780
Electrons travel faster and more effecient in cold conductor. Like the cpu transistors for example, when get cooled down to subzero temps or to extreme temps it materials becomes superconductor of electron/electricity. But even though cpu can perform faster at cold temp it can however has limits.
 

1Tanker

Splendid
Apr 28, 2006
4,645
1
22,780
Electrons travel faster and more effecient in cold conductor. Like the cpu transistors for example, when get cooled down to subzero temps or to extreme temps it materials becomes superconductor of electron/electricity. But even though cpu can perform faster at cold temp it can however has limits.
Yeah, IIRC they used to say that a transistor could double its speed at ~ -120C. Most CPU's can handle this and lower, but most people don't have the equipment/money to continuously keep them at that temp. The AMD FX series chips have a cold bug at ~ -50C. IIRC. :?
 

WR

Distinguished
Jul 18, 2006
603
0
18,980
The reasons an engine has difficulty starting up in cold weather are that 1) ignition spark plugs draw power from batteries which depend on chemical reactions that speed up at higher temperatures, and 2) lubricating oils that allow an engine to run smoothly get more viscous at lower temperatures. And the primary reason an engine would fail at high temperatures is that raw metals expand and contract more noticeably than other materials in an engine, such as alloys, and many engines are designed with a rather limited (but practical) thermal tolerance, beyond which moving parts would experience too much friction to sustain operation.

A car engine is to a CPU as apples are to oranges. There is no "ignition" temperature to reach in a CPU, nor are there supposed to be chemical reactions in an operating CPU, which is a solid-state device (no moving parts). Upper and lower temperature bounds for CPU operation mostly have to do with technical limits imposed by a particular manufacturing processes. I don't know that much about CPU manufacturing, but a much cited example is that recent AMD CPUs don't operate at liquid nitrogen (LN2) temperatures, whereas most o/c records for Intel CPUs continue to be set using LN2.
 

SidVicious

Distinguished
Jan 15, 2002
1,271
0
19,280
isnt silicone plastic type in a way? Wouldnt that contrapt the core making it brittle and shatter internaly? My hypothesis is that if your cpu gets too hot it will crash, same concept if it gets too cold. How about the NANOCOOLER?

Silicone is a polymer which contain some silicon, not to be confused with the substrate of pure (or strained) silicon used to make CPUs...
 

overclockingrocks

Distinguished
Oct 9, 2006
379
0
18,810
Basically on this point, If you cooly your system using commercially available products the cooler i can get the better. Most liquid nitrogen setups drive the system to -10 to -15 at the lowest far from the point when a cold bug hits. You could potentially use nonstandard or industry grade stuff to cool your cpu to the point that it would experience a cold bug but as previously stated most people don't have the time equipment or money to do this, so really the idea of cold bugs is a moot point at this stage of the evolution of pc cooling solutions. I say if you want to overclock go water or for a more extreme version go vapochill case with forced gas cooling neither one will make your cpu unstable due to temps.

overclockingrocks
 
G

Guest

Guest
Euh. liquid nitrogen is MUCH cooler than that. Just with Dry ICe you get at -50/70...

Further, its ability to maintain temperatures far below the freezing point of water as it boils at (77 K, -196 °C or -320 °F) makes it extremely useful in a wide range of applications as an open-cycle refrigerant, including;
(Wikipedia)

And yeah I thought the Cold Bugs comes from the temperature sensor used to shut down if overheated not being design to "understand" -50/150 temps. Am I wrong?
 

CNeufeld

Distinguished
Jun 26, 2006
267
0
18,780
Is it possible for a cpu to run too cold? I keep seeing all these things about cpu's being cooled by hydrogen, helium, dry ice, etc. How is a cpu able to work at -1xx*? I always compared a cpu to a car engine, it runs better when its cooler, but haven't you tried to start an engine in the winter, doesn't work as well. Someone please explain or direct me to another topic or editorial.

I think a bigger concern for most people trying to go cold would be the issue of condensation buildup. Water and computer parts don't mix well together.

Clint
 

sunderlandj

Distinguished
Oct 9, 2006
2
0
18,510
Phase change cooling such as a mach 2 GT, that use an evaporator head achieve -200 and under allowing for amazing overclocks
 

mrmez

Splendid
Agree... unless ure geting close to 0 Kelvin (absolute zero where all molecule vibration [heat] stops), its prob a case of the colder the better. I doubt in ANY home/office environment u can eva have a CPU too cold. Obviously with less than 0c temps there is the prob of condensation when things heat up, but thats another story.
So chill away.

As for engines (which i know much more abt), cold isnt really better, especially in race engines we have a saying 'a hot engine is a happy engine'.

2c
 
you'd need too; this waste heat has to go somewhere, meaning in your face (cooling systems 'extract' heat and send it somehwere else; the second principle of thermodynamics is not strong enough to make waste heat disappear without being noticed)
 

immagikman

Distinguished
Sep 29, 2006
264
0
18,780
I don't believe Silicon is a polymer, since it is basically just sand, it is a glass/crystaline material. Polymers are plastics.

As for super cooling a CPU, you won't see super conducting Silicon nor will the transistors on the chip super conduct, since by definition they are SEMI-Conductors and if they are made to SUPER conduct they will cease to serve their fucntion. The idea of Semi-conductors is that their conductivity is controlled.

What the cooling does is allow you to run the chip at higher speeds without destroying the material due to thermal issues. The faster the CPU is run the more heat it will generate....to the point of destruction, unless adaquate cooling is provided.
 
warning: cooling won't solve the transistor switching speed problem - this is what smaller engraving solves in part.

Otherwise you could get a 8088, heavily cool it then clock it at a terahertz.

Nope, won't work (or would it? the 8088 was made using copper masks and acid for engraving if I'm not mistaken...)
 

immagikman

Distinguished
Sep 29, 2006
264
0
18,780
LoL Yeah there are several issues to overclocking....however I believe at this point, being REASONABLE and not taking things to extremes, the issue of the 90 or 65nm process issues are partly about speed, and partly about heat generation....I think my main point is still vaid :)
 

TRENDING THREADS