Question CPU Temp Sudden Spike To Shutdown Level

[Sertica]

I was browsing Arma servers today when my fan suddenly went max speed. I checked monitors in Asus AI Suite and cpu was over 80 degrees. I hit the power switch in panic, then went into the bios to monitor it without the AI Suite OC. It dropped back normal around 40. I went back into windows and did Intel stress test with the OC settings to see what what would happen. It shot up again, so I restarted, but I’m now unable to enter windows as it says “temperature error”. Now it is overheating also in the bios on default settings. So that machine is unusable until I figure out why this is happening. The fan is running at 2000rpm. The input is 1.6v. The G3258 core is 1.044v. Everything appears normal. I’m stumped.

 

[Supahos]

See if the stock cooler has come loose. It shouldn't wiggle at all. Also confirm that the fan is infact spinning when you try to boot on up

 

[Sertica]

I took the cooler off and put it back on. It does not move. The fan is spinning.

 

[Sertica]

What should be the rpm of a 75mm fan? Google is useless.

 

[Azzyasi]

How sudden is the spike? Temperature can't spike without an explosion or something. So it may be a sensor issue. But if the spike shoots up gradualy in a few seconds then it might be overheating by poor installation of the cooler/no thermal paste.
I'd suggest remounting it and apply new thermal compound.
Also remount the cpu in the socket since it may resolve the sensor issue. (maybe poor contact somewhere)

 

[Sertica]

By that I meant it was constant for 3 years, peaked at 74C in a higher V test the day before, I was in an Arma 3 server minutes earlier, then I heard the fan start to rev abnormally while in the server browser with only a few % useage shown in the process list. So it looked like some serious fault developed just in that few minutes.

I snapped the heat sink on again and it went back to idling at reasonable temp, 47 in the bios and 37 in windows. I must have made it worse when I I did that earlier. The flimsy plastic pegs are tricky. It is still peaking at 88 in IntelBurnTest compared to 67 yesterday with my normal OC. There are holes in the thermal paste that I will fill in later, but something else must be different than yesterday.

 

[Azzyasi]

Reapply thermal paste with either method. Some swear on the dot/line method, I go with full cover thin spread with a card.

Thermal paste degrades in time, and never heard of patching holes... get a tissue and some IPA and clean that stuff off and reapply the TIM compound.

Also take a good look on the pegs and mounting system. Plastic pegs my snap one side and then it just sort of hangs in there. You can go for screw mount (take note if you have a backplate or not)
Some odd mountings used a metal shroud with 4 legs sticking out pressed on the base of the heatsink and that can slip off the heatsink for some reasons.
Another indication that a solid contact is present is that under load, the heatsink gets hot. If the CPU is hot but heatsink is cool/warm then the thermal contact is bad with poor mounting and/or poor TIM application/performance

 

[Sertica]

After I put on new thermal paste and ensured it was fastened well it tested even cooler than before the problem started.