Shutdown due to CPU overtemp... I think?

[gioGats]

Hey guys, looking for some creative troubleshooting help here.

 

I've got a rig I built last November.  i7-6800k cooled by a NZXT x62, Corsair 3000MHz RAM, overclock to 3.9 GHz.  

 

It's been running fine on all range of tasks, including some server tasks that would max out all cores for hours on end.  I ran it for a few hours on a task that maxed out a single core.  Seemed fine.  Then it just shut down.

 

Now when I load bios, it's reporting all sorts of weird frequencies: 2.2 GHz, 2.9, etc.  It's also reporting CPU temp of 100ºC.  I can boot windows but not linux, and even that only lasts for a minute before it shuts down.

 

I removed the overclock, tightened the x62 (it wasn't loose, but worth a try), flashed the bios.  All to no avail.  The outbound tube on the x62 is warm when it's running, so the x62 is carrying some heat at least.  I'm planning to swap in a different cooler tomorrow, but that's not much of a long term solution.

 

What other things can I try to troubleshoot?  What happened?  And how do I make sure this doesn't happen again?

 

 

[LabRat]

the first thing I thought about was a hot cpu. the over clock didn't help. new thermal paste?

I would check the ram too while you're buying stuff.

seek out dust clogs./ make sure fans are running. hook pump to fan header not cpu header.

[Windows7ge]

If you can verify that the CPU isn't actually reaching these temperatures. Like by touching the back of the motherboard (If your case has a cutout) and if it isn't discomforting to keep your hand there then I'd say this is CPU or motherboard malfunction.

 

You also said you can no longer boot Linux. Weather that's a separate drive or just another partition I think that's an indicator something went wrong with your CPU/motherboard. Along with irregular temperatures and frequencies. Either overheat or something is on the verge of death.

[ARikozuM]

Can you check the temp for the motherboard? It could be a bad probe.

[gioGats]

I just tried booting again, to interesting results.  BIOS reported correct CPU frequencies and CPU of 50, rising about 1º per second.  It topped off at 99, and I managed to get a successful boot into Linux to print the attached sensors report.  

 

This time the tubes on the x62 were not warm.  No noticeable difference between in and out tubes at all.  

 

I'm suspecting a thermal paste issue?  Which is concerning because the x62 came with pre-applied paste.  

Mobo was steady at 30º.

[DrMikeNZ]

Which fan header is the x62 connected to on the motherboard?

It would likely be a thermal paste, or the pump might not be working properly. Double check that the power cable for the cooler hasn't been inadvertently unplugged.

[ARikozuM]

9 minutes ago, gioGats said:

snip

Check your paste, tolerance for the block, and the pump itself.

[gioGats]

4 minutes ago, DrMikeNZ said:

Which fan header is the x62 connected to on the motherboard?

It would likely be a thermal paste, or the pump might not be working properly. Double check that the power cable for the cooler hasn't been inadvertently unplugged.

x62 is on the CPU header, and power cables are still in.

 

 

Is there any way to know if thermal paste has gone bad, other than that the problem goes away after reapplying it?  Like a discoloration or texture?

 

[DrMikeNZ]

3 minutes ago, gioGats said:

x62 is on the CPU header, and power cables are still in.

Is there any way to know if thermal paste has gone bad, other than that the problem goes away after reapplying it?  Like a discoloration or texture?

What fan speed is reported for CPU in bios?

You need to take the cooler off to check the paste, and doing so you should reapply paste anyway. Reapply and hope problem goes away, otherwise see if you can get a different cooler to test.

[gioGats]

4 minutes ago, DrMikeNZ said:

What fan speed is reported for CPU in bios?

You need to take the cooler off to check the paste, and doing so you should reapply paste anyway. Reapply and hope problem goes away, otherwise see if you can get a different cooler to test.

Right around what it said in the printout ~750 rpm.

 

This is right around what it had been prior to the issue.