Why is there an instant overheating of the processor from a reboot?

Question

Iron

Collected from purchased, new spare parts sistemnik. There was the following problem at startup (OS not installed yet):

1. I start the system unit, I normally go into BIOS Setup
2. The BIOS Setup shows the temperature of the processor. As soon as I enter the BIOS Setup, it is at about 52-55 degrees and then quickly drops to 45-47, where it continues to stay. I can wait for many minutes - the temperature of the processor does not increase.
3. I change the BIOS settings, I click save and restart. After that, at the POST stage of the boot, one long and three short signals are issued, “CPU Over Temperature Error, Press F1 to Continue” is displayed , I go directly to the BIOS settings - the processor temperature is already at 75-77 degrees, but again quickly drops back to 45.

Those. the problem is that from a reboot, my processor heats up from 45 to 75 degrees in a second, causing an error.

Possible Sources of the Problem

1. Bad thermal paste . First, I smeared the thermal paste that came with the cooler on the CPU. After I immediately received the described error, I removed the cooler and generously smeared everything with ZM-STG2 Super Thermal Grease. The processor itself is pressed tightly to the motherboard by the mount, the cooler to the processor too.

2. Overclocking timings . I did not make any settings for overclocking the computer, as I unpacked the motherboard from the box, I installed it with those settings.

3. The fan on the CPU cooler does not work . I removed the side cover of the case, checked - everything is spinning. I felt the cooler heatsink - it barely heats up (i.e. it feels like it doesn't stay cold, but takes on heat).

4.PSU problem, too high voltage, power surge at start . I don’t know how to check this, the PSU is also completely new.

What other options might there be?

The components that make up the system unit:

1. Processor Intel Core i5 2500K, 3.4 GHz

2. Here is a cooler on it.


3. Motherboard ASUS Sabertooth P67 (rev 3.X) with UEFI AMI BIOS, BIOS version from September 2011

4. Power supply HIPER K800 800W

5. RAM DIMM Samsung 1333MHz DDR3 4Gb x 4 pcs.

6. Video card nVidia GeForce GTX 550Ti, add. power connected

Note:the specified motherboard has a lot of temperature sensors, according to the readings of which it independently regulates the rotation speed of all coolers. So right at the very moment of rebooting, it accelerates the coolers to a loud howl, and then when everything cools down, it works absolutely silently and does not heat up.
Those. this fact confirms that the processor heats up sharply not during operation, but precisely at one moment of reboot.

Answer 1

Try to force the coolers to the maximum number of revolutions. Also, you may have a “buggy” thermal sensor. Once I came across a similar one, they sinned for everything, but it turned out to be the temperature sensor.

Answer 2

I wouldn't be particularly worried about this. Both of my rack servers also start to boot with a loud howl. I think this is 1) testing the fans, 2) calibrating the circuit that regulates the speed of rotation.

Answer 3

5. Bios glitch (namely, temperature determination, it is in order). If at the time of the reboot the speed does not drop to zero, then there is nothing to overheat from. I suggest updating the bios. The site is also 2 weeks old.

Answer 4

Badly smeared thermal paste?
Crooked cooler base and/or crooked CPU cover?

Answer 5

Question on point 5
After looking at the Bios firmware on the Asus website, I found that the latest from 2012/05/02
Sabertooth P67 BIOS 3209
After updating it, its behavior will be interesting.

Answer 6

> plentifully smeared everything
Thermal paste should be applied in a thin layer, solely in order to smooth out irregularities. A thick layer conducts heat worse.

Answer 7

Literally the day before yesterday I fixed this jamb in one of the system units. The problem was, as I thought, in the thermal paste. Replaced it and it worked as it should.
ps: Above we wrote a variant with the acceleration of the cooler on the processor. This is an option, but the Turbo mode did not help me.

Answer 8

Another option is that the lid has peeled off the crystal inside. In new processors, the cover, as far as I read, is made non-removable, and earlier it was possible to remove and replace the dried thermal paste (now it is glue that leads to damage to the crystal when trying to remove it). But, since then everything is in order, then this is most likely the incompatibility of the BIOS with the new stepping, for example, corrected by flashing.
(the fan on the cooler generally does not affect the first 20-50 seconds, the inertia is high. The howl at the moment of switching on can be either due to a jump in the BIOS readings or as a result of not setting the initial values ​​​​for the PWM fan by the BIOS, and after initialization the frequency returns to normal)

Answer 9

I had a similar case, the processes were accidentally overheated (as a result of a detached cooler), after which no matter what I put cooling on it, nothing helped, the processor warmed up to 120 degrees in 10 seconds! Threw away…

Answer 10

Howling is a feature (no one controls the speed of the cooler during the initialization of the periphery), and overheating, if this is not a sensor glitch, may be caused by a processor reset. I don’t know how relevant this is, but in books on P2-P4 it was written that resetting the processor (RST to the leg) causes the circuit of everything that can be closed and gives the maximum current (almost more than from full load). But RST is sent for a very short time, so I don't know whether it is or not.

Answer 11

Thermal paste does not need to be lubricated abundantly. And that is, they smear it with love, just like butter on a sandwich.

Answer 12

And my 5 kopecks, it happened already 2 times, the first time due to abundant thermal paste, coupled with an incorrectly set cooler mount, due to which the cooler did not fit on one side and as a result there was a lot of paste, as a result, heat was very poorly removed when percent heated above 50 degrees, it was decided accordingly, little paste, correct installation.
The second time I had to replace the motherboard with the same one, the defect was not set to 100%, but everyone agreed on the processing of sensor signals or defects in the sensors themselves, as they checked without a cooler, with an infrared thermometer (such with a laser dot), it was there at startup only 52, and the bios showed 78 at that moment.

Answer 13

To begin with, I will explain the work of any cooling.
Firstly, cooling happens, as already noted, 3-4pin.
4pin - has a PWM controller
3pin - does not have a controller, only a cooler speed sensor.
3pin can be controlled through the motherboard, if it can lower the voltage, then the speed will also go down.
When the PC is started, all coolers connected to the motherboard, including the cooler on the video card, start working at maximum speed, then after the BIOS gives them a control signal, they reduce their speed.
An exception may be the cooling of the video card and power supply.
For a video card, everything is controlled only by the video card itself, in rare cases by drivers through the Operating System, so most often "modern" video cards immediately work at 1/3 of their revolutions or a quarter, the block is the same - immediately with the setting.
So it turns out that when you start the PC, your processor cooling first turns on to the maximum, and then the control and speed decrease are already triggered.
In some cases, the cooler may not immediately start spinning at all - this is within the normal range.
Overheating does not happen so quickly, in addition, it also depends on the processor.
New processors are quite technologically advanced and do not work at full power when the PC is started, therefore overheating is not possible.
You can even remove the heat sink altogether and start the PC - for a while everything will work.
The processor will not burn out - this is a high-tech miracle, you still need to try to burn it.
The maximum that can happen is the activation of the "Throttling" protective function, which will reduce the frequency of the processor and force it to slow down and lower the temperature or turn off the motherboard - this already depends on the motherboard.
Now at the expense of the temperature that your motherboard gives out.
Here you already need to look at the problem according to your board model - maybe someone has already discussed it on the forums.
- Check if the "Open Case" mode is enabled, that is, with an open case, without a side cover.
- Check what "Warning" functions are and try to disable.
- Try to turn off the cooler control or set it to work at maximum without lowering the speed and see what happens.
- Check the temperature already when Windows or another OS is running .. It is better to load the processor with Prime95 or some other program that will load it 100%. At the same time, it is better to monitor the temperature through a program that will show a graph, and not just a number, for example, the "Speedfan" program in the last tab allows you to turn on the graph of any sensor and track it, but there are others.
- Try a good thermal paste or at least domestic KPT-8.
Your cooling is normal, but even the standard one is designed for maximum processor load and stable operation.
On it, some even manage to slightly overclock the processor, but of course, good thermal paste and increased cooling speeds are already needed here.