Overheating help PLEASE!

[HappyBuddhaman]

Two days ago, my cpu skyrocketed in temp, went from 33 to 93 in about 10 seconds. Tried shutting it down and restarting it a few times, only to have it rise to 90 upon every boot. Seemed to be my AIO cooling unit, a Thermalrake Extreme Water3.0. No heat in the pipes,and the rad what wasn't hot. Found out that if I shook the read, the temp would come down. Figured the pump blew. Bought a new corsair 100i gxt set it up in a push pull configuration, and experiencing the same thing. The corsair seems to fit loosely on the cpu, which they say is normal. All voltages seem normal. All plugs have been reset. Please help.

Formula VII MB
i4790 cpu
16 GB Trident X RAM overlooked to 2400
Supernova1300 PSU

[Korth]

Some people have commented that the Corsair H100i GTX has a loose backplate. Apparently because different motherboard models have different thicknesses and the mounting posts are too long for thin motherboards, so the cooler can't screw in far enough to sit flush on the motherboard or apply correct pressure to the processor - the entire assembly might be firmly anchored a fraction of an inch too far (and be a useless thermal interface because of the air gap) or it will be loose and wobbly (and, again, be a useless thermal interface).

Anyone who says this is normal is mistaken, lol. The cooler will not work properly unless it is mounted properly. If it doesn't mount properly then it is simply the wrong cooler - Corsair should publish compatibility lists or provide different mounting kits but they don't - so you'll need to mod the hardware (with some metalwork or something) to make it fit perfectly and work right or you'll need to choose a different CPU cooler for your motherboard and processor. Perhaps a measured thickness of thermally conductive pads might work behind the socket, but it's a ghetto solution and probably wouldn't work right for long (if at all) anyhow.

The Corsair H100i (standard version, not the GTX variant) is compatible with your motherboard.

[Malloy71]

Hi! I built a new m8/6700k set-up in early December with the h100i gtx and had absolutely no issues with the backplate being loose, actually I rma'd my first m8 due to a usb issue but I've now had 2 m8's with that cooler on it with no issues and great temps. Are you sure you used the right backing plate screws that came with your board? I believe it comes with 2 or 3 sets of different length screws for installing the backplate.

[trainut]

korth get the h11oi gt if it will fit good ones

[quark54]

I don't know any way a CPU can suddenly skyrocket in a matter of seconds, from 33 to 90 degrees, unless their is a cooling failure, fan failure etc. Having said that, the chances of having two failed coolers like that, one after the other is very small.

Are you sure you are reading the temperature correctly, and that it's not a temperature reporting issue, so software related?

Korth is correct regarding mounting pressure. Mounting pressure directly affects cooling. Precisely why many cooler manufactures deliberately increase mounting pressure above Intels load recommendation by many Newtons.

[cekim]

quark54 wrote:
I don't know any way a CPU can suddenly skyrocket in a matter of seconds, from 33 to 90 degrees, unless their is a cooling failure, fan failure etc. Having said that, the chances of having two failed coolers like that, one after the other is very small.

Are you sure you are reading the temperature correctly, and that it's not a temperature reporting issue, so software related?

Korth is correct regarding mounting pressure. Mounting pressure directly affects cooling. Precisely why many cooler manufactures deliberately increase mounting pressure above Intels load recommendation by many Newtons.

The voltage/power/frequency management on these chips is VERY effective... They can go from nearly room temperature at idle to 80C in seconds depending on frequency and voltage even with functional cooling solutions.

The key is that even with terrible cooling at the lowest settings the 4970k will behave exactly this way. You can turn off the AIO and mine will stay under 35C. With the AIO functioning, it idles at 21-25C. The primary reason is that it is stepping down to 800MHz. When I load it up, it jumps to 4.6GHz in a hurry and I see 55-60C within a single refresh of the temp monitor.

So, odds are, if you have a pump, it stopped. Temps are manageable until you load up the processor. Of course, the OP should definitely check for lose mounting as well, but just know that a failed pump will behave exactly as you described.

[quark54]

Cekim... I don't think you interpreted my post properly.

They can go from nearly room temperature at idle to 80C in seconds depending on frequency and voltage even with functional cooling solutions.

Of course they can. But the OP was obviously referring to a sudden extreme change of temperature without changes in other parameters that would normally be associated with a temperature rise. Precisely why he regarded it as an anomaly. Which was precisely why I commented that it would require a cooling failure.

So, odds are, if you have a pump, it stopped. Temps are manageable until you load up the processor. Of course, the OP should definitely check for lose mounting as well, but just know that a failed pump will behave exactly as you described.

Which is precisely what I implied, a cooling failure. The weirdness is that he had the same behaviour with two consecutive coolers which may point to another issue.

[cekim]

quark54 wrote:
Cekim... I don't think you interpreted my post properly.



Of course they can. But the OP was obviously referring to a sudden extreme change of temperature without changes in other parameters that would normally be associated with a temperature rise. Precisely why he regarded it as an anomaly. Which was precisely why I commented that it would require a cooling failure.



Which is precisely what I implied, a cooling failure. The weirdness is that he had the same behaviour with two consecutive coolers which may point to another issue.

Indeed, I've seen mention that there are "pump" headers using the same part as fan headers but with different pinouts.