Do you use third-party coolers?

[ainwood]

I don't overclock - have no real interest in it. As such, I've always used the stock cooler - and they've generally worked fine.

Over the last week I've been building a new PC, and again just used the stock cooler. Once up and running, I checked temperatures, and they seemed a bit hot: 46°C on low load, and peaking at > 90 on full load (ambient 25°C). So I took the cooler off, carefully cleaned-up the old thermal paste, put some new paste on and reset everything. Idle temperatures hardly changed, but got some benefit for load temperatures, dropping them to a max of 86°C.

To my mind this is too hot. Although its apparently below the thermal trip setting, I see that overclockers target a maximum of about 70 for an overclocked unit, which mine isn't.

So I figured I'd get a new cooler. There wasn't much on offer in my town, but I got a Tuniq one. Attached that, and get much improved performance. On low loads, CPU temps are around 32°C, and at full load (sustained for 10 minutes), it only rose to a maximum of 56°C.

Anyone else have similar problems with the stock coolers?

 

[Genocidicbunny]

It depends on what processor. For instance, the Pentium 4 coolers were notorious for being just barely enough to keep the CPU from melting. Nowadays, the stock coolers have improved, but not the point that Id feel comfortable with one.

The thing about stock coolers is they are often (if not always) mere heatsinks slapped on top of the CPU. They may have a fan blowing hot air over them, but with a modern CPU and the amount of heat it produces, the heatsink just cannot carry the heat away from the contact surface fast enough. At some point, it becomes 'full' and becomes much slower at drawing heat away. This is where you see your 90C temperatures.

Good after-market coolers use heatpipes to actually suck the heat away from the contact surface and to spread it over a much larger cooling surface. They vary widely in terms of efficiency, but one thing can be said: majority of them are better than stock.

I havent used a stock cooler in years, and can't say Im missing out on much. Current rig has a Noctua NH-U12P which keeps my overclocked quadcore at 30C idle and 50C load (or somewhere in the region)

As for why overclockers shoot for 70C. When your temperatures get higher than that, you end up with some funky effects in the silicon after a while. After some time of running at those temperatures, the silicon becomes damaged and you might end up having to replace your cpu. This is even more pronounced these days as the CPU's are built on really small processes. Compare the average 1.5um or 90nm processes of several years ago with todays 45 and 32nm processes.
The thermal trip is reserved for complete heat transfer failure so that your CPU does not literally melt. If you go to Toms Hardware and look up one of their old videos on what happens to CPU's when they have their heatsink removed. The results are quite jarring: 200+C temperatures in seconds. To avoid completely destroying your processor, most motherboards now trip a shutdown if the CPU temp gets to around 100C.

 

[Mike Hussey]

I think for the Core2Duo series, the stock cooler (if installed properly) should net you 35 degrees Celsius at ambient and about 40-45ish under load.

I heard that Turniq thing is an animal though. Can get you around 20 degrees under load.

 

[Padma]

I use a Coleman. It holds 2 six-packs, plus ice.


Oh.



I've always used stock coolers, and never had problems. Of course, I haven't built a system in a few years....

 

[Genocidicbunny]

I think for the Core2Duo series, the stock cooler (if installed properly) should net you 35 degrees Celsius at ambient and about 40-45ish under load.

I heard that Turniq thing is an animal though. Can get you around 20 degrees under load.

More like 40C idle and 65C load. It will vary processor to processor of course, even within the same model line

There's no way you can get a CPU running at below ambient with a regular air cooler. WC its entirely possible, phase change is also possible, but not with an air cooler. Especially on load. You might have been looking at delta instead of absolute temperatures, because delta is just the difference between CPU temp and ambient.

That, and there's no way a Tuniq can outperform a TRUE, which I know cannot get CPU temps down to 20C on idle.

 

[aimeeandbeatles]

I once read about somebody stuffing their computer full of dry ice to cool it down. But that's a bad idea, I'd guess, because I think dry ice has a lot of water vapor. I just know it's a very bad idea.

EDIT: Apparently some people do use it seriously.

 

[Genocidicbunny]

I once read about somebody stuffing their computer full of dry ice to cool it down. But that's a bad idea, I'd guess, because I think dry ice has a lot of water vapor. I just know it's a very bad idea.

EDIT: Apparently some people do use it seriously.

Dear god no. Everything you said here is wrong.

First off, lets begin with "stuffing their computer full of dry ice"

They don't stuff their computer full of it, they fill a copper pot which is attached to the CPU with it.

Secondly, no, dry ice has no water in it at all. If allowed to sublimate in a 0% moisture atmosphere, all you'd end up with is Carbon Dioxide because thats all Dry Ice is, CO2 in a solid form. The water vapor comes from the atmosphere around the cooling pot. There is no other source for it since the liquid used to facilitate heat transfer in the pot is something with a freezing point below ~-80C
In other words, these people arent crazy. Dry Ice cooling is a more extreme form of water cooling, and is usually done on bench rigs meant for serious overclocking. If you want really extreme cooling, you go to LNO or submersive cooling. Either one of these is dangerous for your hardware, and requires a lot of skill (and money) to pull off.

While we're on the subject, and since you will probably end up asking anyways:
LNO cooling -- cooling using Liquid Nitrogen. You do the same thing as with Dry Ice cooling, but you use LNO in the pot. Extremely low temps are possible this way (-100C or so with the processor overclocked and running at full tilt) and it is pretty much the only way you can reach some of the highest speeds possible with a processor.
submersive cooling: You take the whole rig and submerge it in a liquid with a good heat transfer and no electrical conductivity. Mineral Oil is the one often talked about, but it is both expensive and still requires a good case to move the oil around. Your regular case fans will have some serious issues dealing with such a viscous material. There are other chemicals developed for this purpose, but they are often expensive and are only used by people who work with 'supercomputers.' The best example of this is some of the CRAY supercomputers which I believe were submerged completely in freon.

 

[aimeeandbeatles]

Interesting. But a question is, is it dangerous for a computer to get too cold?

 

[Souron]

When first installed my current build, I initially didn't properly attach the CPU fan. So I observed ~90 C temperatures in bios. It looked and seemed secure at the time, so I didn't figure out the problem right away. But After I took it off and put it back on more tightly it worked fine.

 

[J-man]

Interesting. But a question is, is it dangerous for a computer to get too cold?

I don't think it will survive long in liquid nitrogen, that's -196 degrees centigrade.

 

[Genocidicbunny]

Interesting. But a question is, is it dangerous for a computer to get too cold?

I don't think it will survive long in liquid nitrogen, that's -196 degrees centigrade.

As long as there is no mechanical stress on any of the components, they should be just fine. Electronics operate best when its extremely cold.

The main danger is condensation building up around the cooling pot. You would also not want to push or apply pressure to any component close to the cooling pot. At such low temperatures, plastics and metals become brittle. The PCB would likely not be able to take even a tiny amount of flex.

Other than that, it should be just fine. You're not submerging the whole system in your cooling, you're only filling up what's essentially a water cooling pot.

 

[Maniacal]

I've read that liquid nitrogen cooling is not advisable for long-term cooling as it can be pretty rough on its own containment system and if it gets out causes a lot of damage. Has anyone here ever used one?

 

[Genocidicbunny]

Ah, the whole non-like metals thing. Yeah, you want to make sure that your blocks and reservoir are made of the same material. If one is aluminum and the other is copper, you're pretty much creating a chemical batter which will at some point degrade either the block or reservoir to failure.

If you're using the same metal throughout though it should not be an issue.

 

[Cutlass]

I use one of these. When I built my current desktop, the cooler that came with the CPU did not anchor to the mobo correctly, and was unusable.

 

[Genocidicbunny]

Ah the veritable Zalman. They used to be so good, but these days there are much better options out there.

 

[Cutlass]

Get's the job done The only thing I don't like is how bulky it is inside the box.

 

[Genocidicbunny]

My definition of job done must be different. I know they cant keep a modern quad-core running at good temps on load. Heck, even dual cores for that matter.

Friend had one from his old rig. His phenom 2 just overloaded the poor thing. Temps of upwards of 80C on load. A shiny new TRUE fixed that right up though.

 

[Mise]

Sorry to move the thread backwards a bit, but is it even physically possible to get a CPU to ambient air temperature using ambient air to cool it...?

 

[Genocidicbunny]

No. Because the CPU is producing heat, in order to have a net temperature equivalent to ambient temperature, your input needs to be below ambient.

The best I've seen is around 30C, but even that is seriously pushing it.

 

[Cutlass]

It keeps my dual core under 40C on normal conditions.