Sign in with
Sign up | Sign in
Your question

Safe minimum CPU temps

Tags:
Last response: in CPUs
Share
May 23, 2008 6:13:18 AM

Hi all.

So basically here is the situation, my Intel Q6600 is overclocked to 3.2ghz with a 1600mhz fsb and sits under water cooling. My concern is that she idles rather cold ie 9/9/7/7 and we seem to be going into a cold winter. Are these temps OK or should I do something to warm up the chip?

Regards

More about : safe minimum cpu temps

May 23, 2008 6:14:13 AM

Those are insane temps, in a good way. :p  I've seen CPUs in the negatives with phase change cooling, so you're probably fine.
May 23, 2008 6:26:06 AM

JDocs said:
Hi all.

So basically here is the situation, my Intel Q6600 is overclocked to 3.2ghz with a 1600mhz fsb and sits under water cooling. My concern is that she idles rather cold ie 9/9/7/7 and we seem to be going into a cold winter. Are these temps OK or should I do something to warm up the chip?

Regards

I'm sure you are using speedfan to check temps
Use Real Temp
Related resources
a c 125 à CPUs
May 23, 2008 6:27:04 AM

What are you using to check the temps?

The only prob with low temps in this case would be condensation, but i doubt it will be a problem.

Temps can never go below ambient(room) temps. Is the rad outside?

Use core temp and post your temps
http://www.alcpu.com/CoreTemp/
May 23, 2008 6:33:06 AM

@kad: Yip speedfan, CoreTemp doesn't seem to work with my system. I'll try the newest version at home

@Nukemaster, with water cooling it is possible to go below room temp on the CPU as the air is being used to cool off the liquid not the CPU. So if the water can absorb the heat it is possible to go below ambient. Sadly my ambient temps are between 5-15 degrees depending on the time of day. We hit the negatives out here.
May 23, 2008 6:45:12 AM

i think it should be alright. :)  i think so.
May 23, 2008 6:46:21 AM

At least I ain't using a 45nm. The metal in the high-K might contract from the cold when my room temp hits 0 again... (Yes this place is cold!). Then I'd have some serious problems... :p 
May 23, 2008 6:58:28 AM

JDocs said:
@kad: Yip speedfan, CoreTemp doesn't seem to work with my system. I'll try the newest version at home.

The newest is Core Temp 0.99 Betta and it will report higher temps than reality
Real Temp 2.5 is the best
:hello: 

May 23, 2008 7:06:51 AM

Kad, I checked the version of SpeedFan I'm using (with the same CPU) on my old motherboard. The SpeedFan I have agreed with CoreTemp 0.95 which is generally considered correct. So I think those temps are right but I'll check with RealTemp.
a c 125 à CPUs
May 23, 2008 7:33:43 AM

Even water can not go below room temp(unless you are cooling it with something[thermo electric or a compressor]). Water just takes in heat faster then air, but its still dissipated into the room(by way of room temp air passing through the rad). its against the laws of physics to be cooler then its surround environment. Other then that a fridge would not need a compressor to lower its temps it would just use water and a pump.

With you low room temps that would explain everything.

As long as its the latest speedfan it should be fine and accurate(the old ones read 15c too low).

Either way, you should not hurt the chip in any way.
May 23, 2008 7:36:17 AM

JDocs said:
Kad, I checked the version of SpeedFan I'm using (with the same CPU) on my old motherboard. The SpeedFan I have agreed with CoreTemp 0.95 which is generally considered correct. So I think those temps are right but I'll check with RealTemp.

Waiting your responce
:hello: 
May 23, 2008 7:44:11 AM

@Kad, I'll do it tonight (kinda at work).

@Nightfall, ok this is a bit of a simplified explaination as how much heat the liquid can transfer is actually measured in watts which is based on the type of liquid, the radiator and the strength of the pump / pressure. The water itself can only be cooled to ambient temperature(lets assume 30 degrees) and can only take heat up to its thermal limit (in my case I think I'll stick to 60 degrees).

So if the processor is idling and only generating 40 degrees for instance and the water can handle 30 degrees being transferred into it while in the CPU block that only leaves 10 degrees on the CPU (provided of course you have a CPU block that can handle it). The water is now at 60 degrees until it hits the radiator where its cooled back down to ambient (30 degrees in this case). In this case it is possible for the CPU (not the water cooling system) to go below room temperature.
May 23, 2008 8:04:47 AM

So, lil confused here on CoreTemp (Q6600-Tj max 100C) and RealTemp (Q6600-Tj max 95C).

Are the Tjunction Max just programed into the software, or does the software read the Tjunction Max from the CPU?

Cause I get a 5C difference because of that between the 2 programs. :cry: 
May 23, 2008 8:11:24 AM

Grimmy from what I understand they are programmed in, otherwise the problem shouldn't really exist but as far as I know the CPU doesn't actually know its own TJunction.
a b à CPUs
May 23, 2008 8:15:32 AM

The water cannot be heated beyond the temperature of the processor in the same way it cannot be cooled beyond ambient. The CPU ina water cooled system can't drop below the ambient of the location of the radiator without some form of refrigeration.
May 23, 2008 8:16:23 AM

Heh.. I guess the reason why I'm a lil confused, is when they say RealTemp is more accurate, if it's just programed into the software. RealTemp is a nice tool, but someone might as well make a temp tool that has a tj max of 90C to make the temps look even better. :lol: 

May 23, 2008 8:51:02 AM

@milez, not to be nasty or anything against you but if that were the case I could hook up my processor, both of my graphics cards and chipset without the water going over 60 degrees. With that same logic I don't really need a radiator since the water can't go over 80 degrees then.

The water is just a heat transfer mechnism. The more heat it is exposed to the hotter it gets until it reaches either boiling point or saturation point at which point it can't get hotter because it can't absorb more heat.

PS I think some people are miss understanding. I'm not taking my water below ambient, I am however using the water to transfer heat from my CPU to the radiator which drastically reduces the heat sources limitation in relation to the limitation of air cooling.
May 23, 2008 9:11:24 AM

JDocs said:
@Kad, I'll do it tonight (kinda at work).

@Nightfall, ok this is a bit of a simplified explaination as how much heat the liquid can transfer is actually measured in watts which is based on the type of liquid, the radiator and the strength of the pump / pressure. The water itself can only be cooled to ambient temperature(lets assume 30 degrees) and can only take heat up to its thermal limit (in my case I think I'll stick to 60 degrees).

So if the processor is idling and only generating 40 degrees for instance and the water can handle 30 degrees being transferred into it while in the CPU block that only leaves 10 degrees on the CPU (provided of course you have a CPU block that can handle it). The water is now at 60 degrees until it hits the radiator where its cooled back down to ambient (30 degrees in this case). In this case it is possible for the CPU (not the water cooling system) to go below room temperature.

JDocs
The fact is Processor is heat source always generating heat (Non Stop) say 40 degrees
The circulating water is always heated when touching the module seated on processor
When hot water reaches radiator, the air blown at it has the ambient temp.(say 25)
Heat will be transfered from hot water to radiator fins to air till they are equal
Water will go back to processor and get more heat from it till both temps are equal and come back to disipate in air
So secret is in equality
Temp of water can not be more cold than ambient temp
If you have a closed bottle of water on the table in your room with a fan blowing room air on it , Do you think water in the bottle will get more cold than your room temp ??
No of course

The only possible senario is if cooling liquid is not water
If Refrigerant is used running at -10 degrees it is possible, and this is the principale of A/C units
Even using Cold Water cooled up-to 6 degrees will do the job and this is the principale of Central A/C systems in buildings and malls

Grimmy
Someone calibrated SpeedFan following CopuTronix instructions in his sticky and compared the results with RealTemp(which I was very curious to do) and the result was a difference of 1 degree
So you can rely on RealTemp as an accurate program to measure temps.

a b à CPUs
May 23, 2008 9:27:32 AM

@JDoc.

As soon as the water gets hotter than the CPU, the heat will be transferred TO the CPU. when your water cooling reaches it's limit the water will be too hot and so will the rest of your system.
You can't stick a heating element that only reaches 50C into a pan and expect it to boil.
May 23, 2008 9:39:06 AM

Kad, please read what I said above. My water is not below ambient temperature. OK a more complicated explaination. Heat transfer is measured in watts (heat over time).

OK lets try it this way.
Ambient is say 75W.
CPU generates say 150W.
CPU block can transfer 200W.
The radiator removes say 300W (most single radiators can do this) can not go below ambient of 75W.
And the liquid can aborsb a total 200W, start at say 75W for room temperature.

So the liquid, at 75W, hits to block and gets warmed up to 200W when it hits the CPU block leaving 25W on the CPU. Then the liquid travels to the radiator cooling the liquid back down to 75W. The liquid (air cooled) never goes below ambient but the CPU (liquid cooled) sits below ambient.

End result:
Liquid: 75-200W
Ambient: 75W
CPU: 25W

The laws of physics involving air cooling are still entact because the air cooled component is not below ambient but the heat source transfering heat into the air cooled component can be since air cooling laws are only indirectly applied.
May 23, 2008 9:48:09 AM

xtkxhom3r, at 3.2ghz(400x8) I only need 1.3 vCore and 1.5 PPL (haven't tried lowering it furhter yet) however at 3.6(400x9) I need 1.5 vCore and 1.6 PPL to get stable and that adds a lot of heat.

As far as the ambient goes we hit up to 30 degrees during summer and I don't want to have to fiddle with the oc every couple of months.

For the liquid I use Thermaltake coolant which is basically anti-freeze, anti-corrosive (Thankfully... Had an accident...), anti-conduct (Thankfully as well...). Plus they advertise that it can help reduce temperatures over standard options.
May 23, 2008 11:57:27 AM

xtk: I'm just waiting for my SLI board to arrive to put my spare 8800GT into action...
May 23, 2008 1:02:52 PM

JDocs said:
Kad, please read what I said above. My water is not below ambient temperature. OK a more complicated explaination. Heat transfer is measured in watts (heat over time).

OK lets try it this way.
Ambient is say 75W.
CPU generates say 150W.
CPU block can transfer 200W.
The radiator removes say 300W (most single radiators can do this) can not go below ambient of 75W.
And the liquid can aborsb a total 200W, start at say 75W for room temperature.

So the liquid, at 75W, hits to block and gets warmed up to 200W when it hits the CPU block leaving 25W on the CPU. Then the liquid travels to the radiator cooling the liquid back down to 75W. The liquid (air cooled) never goes below ambient but the CPU (liquid cooled) sits below ambient.

End result:
Liquid: 75-200W
Ambient: 75W
CPU: 25W

The laws of physics involving air cooling are still entact because the air cooled component is not below ambient but the heat source transfering heat into the air cooled component can be since air cooling laws are only indirectly applied.


Load of crap. Heat always flow from a warmer object to a cooler object. So in this case your CPU would be heated by surrounding warm air in the case that in turn would be warmed by your room air, bla bla bla. until an equilibrium is attained.

The only way to achieve below room temperature is to use more advance techniques, peltier, closed circuit (like AC units).

So your temperature readings are way out or your system is not a normal Water cooled solution.
May 23, 2008 1:15:35 PM

Well Iafontma those temps are inline with the ambient temperature am experiencing as said eariler firstly.

Secondly its know that water cooling is less effected by room temperatures or at the components being cooled are due to the 2 step cooling.

3rd of all Kad pointed out, its possible if cooling liquid is used and as I pointed out I use Thermaltake's liqud (as stated above) which is advertised as being able to cool systems down below other coolants.

4th if you had bothered to ask you'd know that I have a total of 4 120mm / 80cfm fans and 1 80mm / 60cfm pushing air through my system and 2 120mm / 80cfm + my PSU sucking air out. The air coming out is hardly more than 5-10 degrees warmer than the air going in due to the rate of movement.
May 23, 2008 1:18:43 PM

JDocs said:
@kad: Yip speedfan, CoreTemp doesn't seem to work with my system. I'll try the newest version at home

@Nukemaster, with water cooling it is possible to go below room temp on the CPU as the air is being used to cool off the liquid not the CPU. So if the water can absorb the heat it is possible to go below ambient. Sadly my ambient temps are between 5-15 degrees depending on the time of day. We hit the negatives out here.



This is what you posted previously. You said you were able to cool BELOW room temp. It is impossible with a normal water cooling solution even with special coolant unless your case is completly in some kind of isolating material which in this case you effectively created a fridge.

So you CANT go below room temperature with water cooling. If you ambient air is 5 degres then 7-10 is possible but not BELOW.
May 23, 2008 1:30:01 PM

I said it is possible. Ever heard of a water cooling component called a water chiller?

Secondly as explained both the air is used to cool off the liquid not the CPU hence as long as the CPU block is strong enough and the liquid is capable of absorbing more heat it is therotically possible to cool off anything to 0 degrees using liquid (or water cooling as it is commonly refered to). However you can not go below zero with out lowering the water temperature below zero since at that point something cold is needed to cool down the component as liquid cooling more or less moves heat away rather than deals with it directly leaving the actual work to the radiator.
May 23, 2008 1:43:06 PM

Exactly as i said, not normal water cooling. A water chiller is a type of system that uses the concept of condensing and evaportaing like an AC or other closed circuit.

Most people in the PC watercooling do not use that type of system because below room temperature will most likely create condensation.

And this part : " the liquid is capable of absorbing more heat it is therotically possible to cool off anything to 0 degrees using liquid " would imply that the heatsink is completly isolated from any other heat source. In a normal PC case, the CPU heatsink is not isolated and would absorb any adjacent heat.

So normal watercooling with only heat sink,radiator, fand and water liquid will not go below room temperature
a b à CPUs
May 23, 2008 2:17:41 PM

I'm with lafontma here. And you never mentioned a chiller!
a c 125 à CPUs
May 23, 2008 3:52:24 PM

mi1ez said:
@JDoc.

As soon as the water gets hotter than the CPU, the heat will be transferred TO the CPU. when your water cooling reaches it's limit the water will be too hot and so will the rest of your system.
You can't stick a heating element that only reaches 50C into a pan and expect it to boil.

Dead on.

EDIT
So now you say you are running a Chiller. Well guess what? thats not water cooling alone.

The chiller is giving the water a below ambient temp to work with. Its not some kind of magic of whipping water around so much that it gets cooler then ambient.
May 23, 2008 5:07:03 PM

Nukemaster Did I f*@!ing well ask if he had heard about it or say I had it?

I run a DTek Fuzion, Thermaltake P500 and dual radiator with dual 120mm/80cfrm fans. I have already said that we hit 5 degrees out here and idling at 400x6 (I keep SpeedStep enabled) it is possible to hit those temps. I didn't say that I go below ambient only that its possible.

If you read what I've been it its that if the water has the capacity to absorb the heat that a component is generating and that its block can transfer the heat its possible. I didn't say that I'm doing it.
a c 125 à CPUs
May 23, 2008 5:26:59 PM

No need to get so mad....

With all your posts you did make it look like you where doing that(cooling below room temp with water alone, I already said with your low room temps it was more then possible to get that cool. Its just all the water lower then ambient you kept pushing was what made people say WTF). I also clearly said that it can not be done without extra equipment(lower then room temp).

Its not like you said it can be done with a chiller or other cooling method combined with water.

Either way you have your answer. No harm will be done to your cpu.

We run a computer in the garage at -30c(winter) and lower. I just had to disable the bios temp warning and run it all the time to keep it from getting condensation from the on/off cycles.
May 23, 2008 5:36:50 PM

Nukemaster, sorry man its not just you. Its like saying a car can do mach 1. Its been done. Its not like its a normal car but we don't say that non-standard car did it just that a car did it, until we go into detail.

Other than chillers the liquid is rather import to. The Thermaltake stuff I'm using (Got a spare bottle) is always cold to the touch. Its a liquid component is from refridgerators.
May 23, 2008 5:40:45 PM

Where do you live? Argentina?

edit : nvm , i just saw south africa.
May 23, 2008 5:44:31 PM

Dariushro The area I live in rather commonly hits negatives. Try this with load shedding (also called pre-emptive power cuts)... No heaters or lights...

Edit: Negatives during winter...
a c 125 à CPUs
May 23, 2008 6:02:26 PM

JDocs said:
Nukemaster, sorry man its not just you. Its like saying a car can do mach 1. Its been done. Its not like its a normal car but we don't say that non-standard car did it just that a car did it, until we go into detail.

Thrust SSC was the first i think(2 jet engines at the bonneville salt flats)


JDocs said:
Other than chillers the liquid is rather import to. The Thermaltake stuff I'm using (Got a spare bottle) is always cold to the touch. Its a liquid component is from refridgerators.

The more heat it can dissipate the better.

With your temps, I would leave the computer all the time because if it gets too cold when it warms up next time you may get some condensation. Also don't bother cooling down the hard drive(s) since Google posted some info(taken from all the drives they ever had die in there servers) saying below a certain temp hard drive failure is more likely.

http://www.engadget.com/2007/02/18/massive-google-hard-...

There is a PDF for download. Dry reading, but too cool may be bad for a drive.
May 23, 2008 6:25:46 PM

Well yeah perhaps I should remove the fan blowin air directly over them...

The standard failure rate for my drives is about 0.34% so I'm hoping I'll be fine.
May 25, 2008 3:07:12 PM

RealTemp 2.5 and I are disgreeing. As far as I understand if your processor is idling (at stock speeds / Intel stock HSF, I'm doing stock speeds / water cooling) it should be only about 3 degrees above the ambient temperature. However RealTemp is reporting 13 degrees above ambient temperature, strangely it is detecting my processor's TJunction as 95C instead of 105C. Other than that it is extactly 10 degrees higher than SpeedFan.

Perhaps RealTemp is detecting my Q6600 G0 as a Q6600 B3?
a c 125 à CPUs
May 25, 2008 4:07:37 PM

Real temp always does that. i am not aware of any cpu with a 105 tjmax(most are 85 and 100[or 95 for real temp])

Since Intel has never published such info(they do not use tjmax on desktop cpu's they use a max case temp instead for there thermal spec) the software dev's are kind of guessing at it.
a b à CPUs
May 25, 2008 10:31:05 PM

Haha, I glad to see i'm not the only one to put my system thru freezing conditions, I had my heater break down winter of 06, And the temps were in the negatives in F°. I remember noticing my computer was quieter than normal....I looked in my window and my processor fan was not spinning, Everest was reporting my old X2 3800 running @ 12c.

I fired up a game to make sure it wasn't broke lol. I froze my ass of that winter, with only a Heater fan to warm up my room.
May 26, 2008 6:48:06 AM

@Nukemaster, yes Intel didn't publish the tjunction for the Q6600 but they have for many dual cores. The problem lies in the nature of the Q6600 being 2 E6600s stuck together. Top of the E6600 are boxed as the E6700 and used to construct the Q6700. Easy to give a tjunction for that since only the top chips are being used. The Q6600 quality varies a bit more. So the one E6600 could have a tjunction of 95c while the other has a tjunction of 105 on the same processor. Which do you market it as? Go for 95c and people will complain that your underrating the chips. Go for 105c and suddenly your lying. Keep your mouth shut and you avoid the problem. This also means that each Q6600's tjunction is unique, hence no official statement on its junction value. This also explains why many people with dual cores report a 5 degree difference between the sets of core. One has a lower tjunction which is used to calculate the temperature of the cores making the die with the higher tjunction seem colder when it actually isn't 5 degrees colder but has a 5 degree higher thermal limit causing a miscalculation.

On my processor for instance the first E6600 overclockes like a dream. I can do Prim95 small FFT 24/7 at 3.6ghz off just 1.4 volts however the second one requires 1.47 to be stable at the same level. However luckily the temps remain roughly in sync.

To figure out what your tjunction is is fairly simple, load up RealTemp and check what it says with the processor at stock (95c tjunction) and let the processor idle. On stock cooling at stock speeds the processor (according to Intel) should idle about 3c above ambient. If your tjunction differs from RealTemp's tjunction it will normally be by 5 or 10 degrees. Hell the first E6600's tjunction may even differ from the second ones.

So going by the fact that at stock speeds under water cooling it was idling 13c above ambient according to RealTemp I can safely assume that my jtunction was 10 degrees higher. Since your core temperature is calculated from jtunction - dts +10 to the tjunction puts the temps spot on with what was expected.
a c 125 à CPUs
May 26, 2008 3:41:53 PM

I had uneven temps with one heatsink even when remounted, but with a freezer 7 pro the temps stayed almost dead in line(1c offset sometimes.)

As long as my cpu does not throttle i am not worried about the temps, but that is something to look into(the different TJmax's from different cpus even when they are the same model.)
May 26, 2008 4:57:40 PM

Kinda what I just said but I was aiming more at the quads since they have to deal with dual tjunctions. Luckily we have the idling / ambient info for calibration.
!