CPU overheats in less than a minute

[Patton]

System:
CPU: AMD X2 4850e 45-watt dual core CPU
MB: Gigabyte GA-MA78GM-S2H rev 1.0
HSF: OCZ Vendetta 2


When I turn on my computer, the CPU temp rises until the computer turns itself off, which usually takes less than a minute. The temps of each core (according to CoreTemp) keep rising until they are both over 100C, at which point the computer shuts itself down. Readings from Hardware Monitor and Speedfan also show core and CPU temps over 100C, however HDD and System temps are normal (30's). This all started happening out of the blue (everything was fine one day, no hardware or software changes, then this started happening).

I thought maybe there was poor contact between the CPU and heatsink, or the heatsink had become damaged in some way. I checked the heatsink for damage or bad HSF-CPU contact (none, good contact), re-seated it twice with TIM, and made sure of good contact between the heatsink and CPU. However the CPU still overheats in less than a minute.

Also, when the temp readings are over 100C, the heatsink does not even feel warm to the touch, so I'm not sure if the CPU is actually overheating.

From the time when the computer worked fine to when it now quickly overheats, no changes of any kind were made. It just worked fine one day, and overheats in less than a minute the next.

Has the CPU possibly gone bad, or could the problem be the motherboard, or even the power supply?

 

[kyeana]

if the computer shuts itself down then its probably the processor (and speedfan seems to confirm this). How much TIM did you use when reseating the heatsink?

 

[Mondoman]

Does the CPU cooler fan work?

 

[fleakiller]

Does the CPU cooler fan work?

+1 and or voltage

 

[Guest]

try a differnet fan for your processor and check out to see if you can buy a different heatsink. Maybe on made for the processor. I have a dual core also and I had to get a different fan because im fans start slowing down cause my processor to overheat. you should read 30's- 40's for you processor, 50's
top. 60's you need a different strategy.

Intel Extreme 9000x 3.0Ghz @2 cores, 4GB 667Mhz Ram, 500GB @ 7200RPM +2-8800GTX SLI, 1920x1200. keep that stuff cool'd
3d mark 06 - 13k

 

[blackwidow_rsa]

Bios update might correct those. If you tried everything, RMA it

 

[Patton]

+1 and or voltage

Nothing changed from when it worked fine a few days ago, to when it now overheats in less than a minute. The voltage wasn't increased, and the fan on the Vendetta 2 heatsink works fine.

I thought maybe the contact between the HSF and CPU wasn't good, so I re-seated the Vendetta 2 twice with new TIM each time, but I keep getting the same results (drastic overheating, computer shuts down).

Would a problem like this point to a bad CPU, or something else (motherboard, power supply).

 

[Patton]

I decreased the voltage from 1.25v to 1.15v, and the temperature readings did go down. However they are still way higher than they should be. Also, I had re-seated the heatsink twice and made sure of good contact, so that isn't the issue.

idling @ 1.15 volts: core #0 is 58 C, core #1 is 43, CPU temp is 75 C

Does this seem more like a problem with the CPU or the motherboard?

 

[fleakiller]

when you deseated the heatsink how did the tim look on the cpu was it thin like the heat sink was sitting all the way down or was it thick?

 

[Patton]

when you deseated the heatsink how did the tim look on the cpu was it thin like the heat sink was sitting all the way down or was it thick?

The TIM was thin and spread out, and the heatsink had been in good contact with the CPU.


Does the CPU itself have temp sensors, or does only the motherboard have temp sensors.

I ask because when programs like CoreTemp, Everest, Speedfan, and Hardware Monitor are all reporting temps over 100c for the CPU, the heatsink does not even feel warm to the touch (and I know there's good CPU contact).

Would this point to bad CPU sensors, bad motherboard sensors, or something else.

 

[Mondoman]

Only the CPU has the CPU temp sensors. Your problem is pretty clearly that the CPU cooler is not working properly. Either its fan is not working right or the heat sink itself is not working right (e.g. heat pipes leaked or came loose from sink). Every time you let your CPU heat up to the auto-shutdown point, you are damaging it. Replace the CPU cooler before you try running the system again.

 

[Patton]

Only the CPU has the CPU temp sensors. Your problem is pretty clearly that the CPU cooler is not working properly. Either its fan is not working right or the heat sink itself is not working right (e.g. heat pipes leaked or came loose from sink). Every time you let your CPU heat up to the auto-shutdown point, you are damaging it. Replace the CPU cooler before you try running the system again.

Does the CPU sensor monitor CPU temp only, or does it measure core temps too (core 0, core 1). I'm getting drastically high temp readings for not just the CPU, but for both cores. If the CPU sensor only monitors CPU temp but not cores, the problem may be something other than the CPU.

The heatsink (OCZ Vendetta 2) and fan is working fine. I tried using another heatsink, and had the same (bad) results. With the temp readings this high, I tried touching the heatpipes just above the CPU contact point, and they were barely warm.

These are my readings from different programs:

@1.10 CPU voltage:

Hardware Monitor:
Temp0: 34c (don't know what this is)
Temp1: 73c (don't know what this is)
Temp2: 90c (don't know what this is)
Core #0: 43c
Core #1: 58c
HDD: 36c

Core Temp:
Core #0: 58c
Core #1: 43c

Everest:
Motherboard: 73c
CPU: 90c
Core #1: 58c
Core #2: 43c
Aux: 34c

Gigabyte EasyTune5 Pro:
System: 34c
CPU: 73c

I am starting to think it's a problem with the motherboard, but I don't know how to isolate this problem.

 

[Mondoman]

Touching the CPU cooler tells you nothing, because you can't touch the cooler/CPU interface where the heat is being transferred. Trying another CPU cooler was a good idea; since that didn't work (assuming it has enough cooling capacity and was installed correctly), that points to the CPU itself or MB. To see if MB speed control of the CPU cooler fan is the problem, plug the CPU cooler's fan into a fan header other than the standard CPU FAN one. This should let the CPU cooler fan run at max speed continuously.
OTOH, if you've been overclocking the CPU, you might have damaged the CPU and that might be the problem. The way to test that is to try a different CPU.

 

[Mathos]

Is your computer in a tower case? It could be an issue of the Cooler appears to be well seated, but once the case is stood upright it shifts slightly or leans a bit. That could be caused by a bent or faulty retention spring, or the molding around the processor could of worked itself loose. I myself had one of the little lugs break off one side of the retention ring a while back.

I'd ask if there was some way a pin on the bottom of the processor got bent and is causing a short?

That 45w cpu should be able to be cooled almost passively.

 

[Grimmy]

Looks to me that the HSF installation went bad.

I basically agree with Mathos.

Things I'd try:

Lay the PC case on it side to even the weight pressure on the CPU. If the temps get cooler, that might be an indication the retention brack may be broken, or have a crack in it.

If the above is true, I'd take the MB out and check the back to see if the mounting retention bracket is broken from the back.