42 réponses à ce sujet

[Badly Lit]

Your CPU is idling at 56C? That's way too hot... Have you tried reseating your heatsink? Sounds like you applied too much thermal paste or didn't put the heatsink on correctly.

[FFMagee]

My mobo has 2 x16s and an x8. However, I distinctly remember hearing about not wanting to use the second x16 unless I'm doing SLI.

[Badly Lit]

You can use either slot; there won't be any negative effects unless it's an x8 slot. Even then, the only effects would be slightly less bandwidth. Unless you have an extremely high end card (such as a 5970), an x8 slot won't hurt your performance at all.

[Valaskjalf]

56C is very hot for idle. My proc idles in the 30's and maxes out at 60C with dragon age (on the stock intel cooler)

[StrikeSaber47]

Valaskjalf wrote...

56C is very hot for idle. My proc idles in the 30's and maxes out at 60C with dragon age (on the stock intel cooler)

Pretty much the same here except I was running cool and well with an OC of 2.61 GHz on stock cooling. I think the Intel Speedstep was involved in running on lower clock frequencies when I am not gaming, and when I do well it doesn't seem to surpass 70C the most. My GPU on the other hand... Anyways, I think you got some serious problems with your cooling setup. Most Core 2 Duo's surpassing 3 GHz (with a FSB over 325+ MHz x 4) needs aftermarket cooling and paste to run cool. So lower the clocks or get abetter cooling system or try to see what is going on with the current one.

[FFMagee]

Well, I've determined that the strange fan noise that I've heard is coming from the heat sink. Obviously, a fan issue in the heat sink is related to my high temperatures. I'm essentially snowed in now, so I'm going to let my computer take it easy, so to speak, and take everything apart this weekend after I get some more thermal paste. Thanks for all of the advice and feedback, guys.

Modifié par FFMagee, 08 janvier 2010 - 12:04 .

[Azimuth0001]

more information is needed to help with this.



Motherboard?

Ram?

Overclocking at all?

What have you tried already? (increase fan speed in bios? Lower clock speeds?, verify thermocouple and fan operation?)

Which exact cpu do you have?



can you post some charting data showing it is indeed a cooling issue, and not a ram or power issue?



Speaking of power, what kind of PSU are you using? Your video card uses a lot of power (not as much as some, but it's not exactly eco-friendly). If your PSU is anywhere close to it's limits, then the voltage will drop considerably, and fast. When this happens components draw more current, and since current is what generates the heat, you end up with odd heating issues that can be a pain in the ass to replicate.

[FFMagee]

Edit: Would have deleted this post if possible.

Modifié par FFMagee, 08 janvier 2010 - 01:34 .

[Franpa]

Yes, we need to know your exact PSU as well as the exact Video Card model to ensure your PSU can power everything adequately.



To cross out the video card as the noisy fan you hear, install Rivatuner and increase the fan speed to the maximum. Does it sound like what you hear while playing a game? If not then your video card fan is not what is generating the noise, you can set your video cards fan back to the previous setting (usually Auto).



Valaskjalf, what is your processor? Also for other Intel i7 920 people, what temperature do you idle at and operate at?

[lizardglenn]

FFMagee
have a little look at this CPU cooler  test roundup even the Zalman CNPS9900 LED which is the weakest cooler there is getting ok temps with an i7 920 @ 3.76Ghz getting stressed with linpack
http://www.xbitlabs....p_18.html#sect0
also look at the bit under the last pic on this page that has a little about Maximum allowed temperature on an i7
http://www.xbitlabs....ng_3.html#sect0
PS: not telling anyone to OC it is just handy to look at info like this

Modifié par lizardglenn, 08 janvier 2010 - 04:38 .

[FFMagee]

I ran Rivatuner, and it did not make the noise when at 100%.  I'm not even sure that my gfx card ever got to 100%.  That sucker was loud, but did not make the noise I heard.   The reason why I had determined it to be the heat sink is it happened again as I was running Prime95 through RealTemp.  I'm not currently overclocking.  If you consider pushing the RAM speed past the mobo default (but within the RAM's spec) to overclocking, then I was temporarily.  However, setting the RAM back to the mobo default of 1066 only temporarily gave me a lower temperature.  That's how I was getting that 56C idle.  I have not gotten it since then. 

My graphics card is BF Tech's 280 GTX OC 1GB - Link Here
My PSU is Antec's EarthWatts EA750 ATX12V - Link Here
My RAM is Corsair's 6GB (2 x 3GB) DDR31600 - Link Here
I have an Intel i7 920 - I believe this is the link, but the memory spec seems to be inaccurate, based off this news 

The noise is not constant.  However, if there is a failure with the fan in the heat sink, it would certainly make sense why temperatures are so high.

[FFMagee]

Looking at instructions from Arctic Silver (who I believe supplied the compound that came with my V8), it would appear that my issue may be stemming from improper distribution of the thermal compound. Arctic Silver recommends a thin strip directly over where the cores should be. However, without any instructions of that nature with the compound we had, we did the traditional method of a pea sized drop spread across the entire heat spreader.

[Gorath Alpha]

FFMagee wrote...

Looking at instructions from Arctic Silver (who I believe supplied the compound that came with my V8), it would appear that my issue may be stemming from improper distribution of the thermal compound. Arctic Silver recommends a thin strip directly over where the cores should be. However, without any instructions of that nature with the compound we had, we did the traditional method of a pea sized drop spread across the entire heat spreader.

Yes, that would be ten times too thick! 

The old descxription as a "rice grain" size drop, and depending on the viscosity and the kind of clamping pressure, the instructions might even have said let the heat sink base do its own spreading from there (center) by itself. 

Actic Silver comes with a quite large and detailed instruction booklet, including directions for heavy burn-in to get it to "set" properly.  It's a decent thermal compound when its full properties are understood, especially if you overclock CPUs seriously, but it's just not suitable for ordinary day to day usage (at least not without a serious understanding of how to use it correctly)! 

Gorath

[Valaskjalf]

Franpa wrote...
Valaskjalf, what is your processor? Also for other Intel i7 920 people, what temperature do you idle at and operate at?

Core 2 Duo E8500 6MB L2 @ 3.16 ghz 65W  - Intel stock cooler (it does look taller than the one for the E7400)

[FFMagee]

Well, file this under "You can't make this **** up." I'm suprised that my computer was stable at all. I imagine that has a bit to do with the 15 cm fan on the side of my case.



- The plastic film on the bottom of the heat sink was still attached.

- The thermal compound was spread as I described earlier, but had expanded a bit more (my guess by heat sink pressure or just by use) outside of the processor. The excess being off the processor probably didn't hurt much, but still.

- The heat sink fan was blowing in the wrong direction. As opposed to blowing the hot air out of the case, it was blowing it further in.

- Probably directly related to the above issue, the fan in the heat sink was DEAD. It was a sheet bearing, according to my brother, and he said that it had gotten "all gummed up." So we've since replaced the fan and pointed it out of the case.



With all of these fixes (proper thermal compound application, no plastic film on the bottom of the sink, new fan in the correct direction), I'm getting much better temps. With a handful of programs (Google Chrome, a Twitter Program, FeedDemon, my Blackberry Desktop Manager, and like 3 other things) my PC is idling at 36C. At maximum (since running Real Temp), the hottest core got 42C. I'm still in the "break-in" process for the thermal compound, so if anyone could walk me through how "thermal cycle." I would be most appreciative. The only thing I could really figure out was that I'm going to want actually turn off my computer for the next couple of days.





Once again, thanks a lot for all of your advice and help. Words can't describe the weight off my shoulders now that I don't have to worry about the fact that I'm running my computer at dangerous temps. =)

[Magic Zarim]

Franpa wrote...

Gorath Alpha wrote...

CPUs and stock coolers are desigbned to run for years and years at whatever utilization they reach, be it 10% or 100%. That's a fact. Any system unable to do so has a cooling problem, not a software problem. That's the way it has always been, and it's not going to change.

Intel said recently that there heat sinks for all there CPU's are designed for normal operational situations, not for burn-in tests such as CPU benchmarking. A CPU benchmark is typically designed to stress the entire CPU and all it's cores (if applicable) with both a constant and consistent work load.

You can stress your CPU, but I strongly wouldn't reccomend letting it operate like that for hours.

Just FYI, Prime95 and such programs ARE considered normal operation. Prime95 is nothing more than a distributed computing client to well.. discover prime numbers. Normal operating conditions refer to the physical environment like ambient temperatures, applied voltages and airflow. Not the type of software running on it.

[Gorath Alpha]

The direction that an axial flow fan is pointed should be based on whether the CPU socket is "crowded" or not (and whether a cooling duct is included).  When there are a number of heat-producing parts all crowded around, the best direction is "DOWN" and through the heat sink's fins, in order to spread out to the other parts nearby. 

Then, a cool-air duct, which I refer to as a "Schnorkel" extends (often) from the top of the fan to the outside of the case through a grill in the access panel.  When you have that ductwork in place, an UPWARD blowing fan is not taking proper advantage of the cooler air availability. 

Some of the most popular of high performance CPU coolers are of the "Tower" design, with a parallel air flow fan on the side of the heat sink, and those should be pointed toward the rear panel and any exhaust fan mounted in the rear, if you have the space to do so (a couple of those I've tried fit poorly in MBs that have a "crowded socket" design, and had to be aimed toward the top of the PC case). 

Gorath
-

[FFMagee]

Holy ****. Running Prime95, my maximum temperature is a good 10 degrees colder than my idle temperature when I was running my surprisingly stable machine.

Gorath Alpha wrote...

The direction that an axial flow fan is pointed should be based on whether the CPU socket is "crowded" or not (and whether a cooling duct is included).  When there are a number of heat-producing parts all crowded around, the best direction is "DOWN" and through the heat sink's fins, in order to spread out to the other parts nearby.  

Then, a cool-air duct, which I refer to as a "Schnorkel" extends (often) from the top of the fan to the outside of the case through a grill in the access panel.  When you have that ductwork in place, an UPWARD blowing fan is not taking proper advantage of the cooler air availability.  

Some of the most popular of high performance CPU coolers are of the "Tower" design, with a parallel air flow fan on the side of the heat sink, and those should be pointed toward the rear panel and any exhaust fan mounted in the rear, if you have the space to do so (a couple of those I've tried fit poorly in MBs that have a "crowded socket" design, and had to be aimed toward the top of the PC case).  

Gorath
-

My processor is placed rather close to the motherboard-height output fan on my case.  Every other fan (but one) is an input fan, so it seemed a bad idea to have the Heat Sink fan blowing against everything else and keeping its heat in the case, even if it's just a little bit longer.

Modifié par FFMagee, 09 janvier 2010 - 10:16 .