Overkill cooling - how cold is too cold

Status
Not open for further replies.

king_maliken

Distinguished
Jun 12, 2011
137
0
18,690
Hey guys,
well I successfully made a cheap way of making liquid nitrogen and then starting thinking of doing a cooling system for my PC using either extremely cooled gas nitrogen or liquid nitrogen. Now I was wondering, how cold is too cold? I don't want to damage my rig, I just recently bought it.

Notes: I currently just have a fully water cooled system (pretty much all custom made) and system temps are fine, I just want to do something overkill.
i7-2600k OC 4.8
XFX 2x6970 OC 1000/1475
ASUS P8P67 WS REVOLUTION
G.SKILL Ripjaws X DDR3 2133
OCZ ZX Series 1250W
Vertex 3 MAX IOPS 240GB
WD 2TB Caviar Black
 
Solution
That depends of your CPU.

Intel CPUs have coldbug, that means that you need find the "ideal" temp where the CPU works extremely good at low temp (-170ºC / -180ºC) while AMD can go -190ºC without problems. Liquid Helium is the top for extreme overclock a sub-zero temps but it's not very used because the temps difference between LN2 and Helium isn't big.

I'd not say "overkill" since extreme cooling with dyce, SS, cascade or LN2 are used only for benchmark purposes and not 24/7, Phase change can be used for 24/7 but with that system you can not go below -50ºC or -55ºC and the power use is high.

SB with LN2 exist, you only need check the correct page: www.hwbot.org
Lots of benchmarking folks use LN2 for benching, but it isn't that viable as a long-term cooling solution unless you have a way to continuously get it into the LN2 pots or use it for a super-chilled WC loop.

It's possible to get too cold, as you sometimes need stuff like hair dryers to get the CPU to post when using such low temps.
 
Well I know that the standard transistors that intel used before the 3-D transistors, were ok with extreme cooling, but with the new transistors could the material properties of the silicon-on-insulator substrate change enough to hinder normal transistor operation? I haven't seen or heard anyone using LN2 for the sandybridge.

~currently building another LN2 generator, to supply the system constantly to support long term cooling. With proper design i should be able to avoid most of the condensation (hopefully anyhow).

If a straight liquid nitrogen cooling is not viable, I'll just do a solid water cooled system, which would be cooled by LN2. Possibly not use water but a gel of some sorts...

EDIT: I should add, this would be for GPU, CPU, Northbridge, Phase transistors and anything else that gets hot.
 
That depends of your CPU.

Intel CPUs have coldbug, that means that you need find the "ideal" temp where the CPU works extremely good at low temp (-170ºC / -180ºC) while AMD can go -190ºC without problems. Liquid Helium is the top for extreme overclock a sub-zero temps but it's not very used because the temps difference between LN2 and Helium isn't big.

I'd not say "overkill" since extreme cooling with dyce, SS, cascade or LN2 are used only for benchmark purposes and not 24/7, Phase change can be used for 24/7 but with that system you can not go below -50ºC or -55ºC and the power use is high.

SB with LN2 exist, you only need check the correct page: www.hwbot.org
 
Solution
I'm not big on just one time benchmarking... I'm looking for a 24/7 cooling option, I don't care about power consumption at all, since I produce my own. My goal was simply to make something with my new LN2 generator/s, because for my purpose alone, I do not need that much computing power. I guess I'll keep building model rockets then, thx saint19.
 
The problem with sub-ambient cooling is condensation.

Put a glass of Ice water next to your computer, and notice in a few minutes there will be moisture on it, and in 20 min there will be a small puddle of water on your desk. That same thing will happen in your computer.

So now you have to insulate all your cooling surfaces, it’s just a pain.

 
Status
Not open for further replies.