Computer throttling and switching off randomly

[r0uge]

Hi all, I'm having some issues which I believe is caused by my CPU and heatsink. This includes throttling issues for a long time now and recently random shut downs and I don't know how to fix them.

My CPU will drop from 4GHz to 1.4GHz randomly while gaming, it's not overclocked or anything. My CPU fan is running at max speed at all times (I have this set in the BIOS).

When gaming or under load my processor sits at about 81 degrees celsius, and sometimes goes up towards the 85 mark.
I have the ARCTIC Freezer Xtreme Rev. 2 CPU heatsink, however it will not spin past 1486 RPM at full load. It is supposed to be a good heatsink to my knowledge but something might be wrong.
I am not able to change the fan speed using SpeedFan and it will not go faster than 1486 RPM, it is set to 100% speed in the BIOS. I have one rear chassis fan which spins at full speed at about 1130 RPM.

In the last 2 days my computer has switched off randomly without warning, not like when I shut it down but rather as if the power was unplugged. Please help me fix this issue especially and the thermal throttling too if you can help me with that as well.

Here are my specs:
AMD FX-8350 CPU @ 4GHz stock clocks / voltage
Nvidia GTX 980 GPU @ stock clocks / voltage
ARCTIC Freezer Xtreme Rev. 2 CPU Heatsink
16GB Corsair Vengeance RAM
Corsair CS650M PSU @ 650W
ASUS M5A78L-M/USB3 Mobo

Thank you for your time

 

[GhislainG]

Your quiet cooler may be the issue, but the cheap motherboard can also cause throttling, particularly when using a tower style cooler. Stress test your CPU and post the thermal margins using AMD Overdrive; do they drop below 10°C? Does the CPU frequency drop to 1.4 GHz?

 

[r0uge]

During the stress test the Thermal Margin dropped to -11.6 degrees celcius, and the cores often throttled down from 4018 MHz to 1406 MHz, from 20X multiplier to 7X multiplier and from 1.3375 V to 0.8875 V

 

[Alan Caldwell]

Check which mobo it is.

 

[GhislainG]

During the stress test the Thermal Margin dropped to -11.6 degrees celcius, and the cores often throttled down from 4018 MHz to 1406 MHz, from 20X multiplier to 7X multiplier and from 1.3375 V to 0.8875 V

Do they drop gradually or all of a sudden from positive to negative? As soon as the thermal margins get close to 0°C, the frequency drops to 1.4 GHz (that's what throttling does to prevent the CPU or the VRM to overheat).

 

[r0uge]

Mobo is ASUS M5A78L-M/USB3.
When running the stress test the Thermal Margin drops very quickly to -11 degrees. The cores are not constantly throttling, they jump between 4GHz and 1.4GHz.

 

[GhislainG]

Mobo is ASUS M5A78L-M/USB3.
When running the stress test the Thermal Margin drops very quickly to -11 degrees. The cores are not constantly throttling, they jump between 4GHz and 1.4GHz.

If they gradually drop to -11°C, then your cooler isn't good enough for that CPU; if they very quickly drop from more than 10°C to -11°C, then the VRM causes the throttling. To be honest I'd suspect the crappy motherboard with the non-heatsinked 4+1 phase VRM to cause the issue. Using a tower style aftermarket cooler only makes it worse. If your case is ATX, then you should replace the motherboard with a better one like the new Assu 970 Pro Gaming, the MSI 970 Gaming, the Gigabyte GA-970A-UD3P or a more expensive 990FX motherboard like the Asus M5A99FX PRO R2.0, etc. You need a very good VRM to prevent throttling when using that CPU.

 

[r0uge]

I'm more concerned about my computer shutting off randomly for the moment and don't have the money to splash out on new parts. Do you have a solution for the problem itself?

 

[Alan Caldwell]

You do need a better mobo and a PSU IMO.

 

[GhislainG]

A cheap motherboard is designed around the stock cooler that spins very fast and therefore helps cooling the VRM (your quiet cooler blows no air towards the VRM). An easy and free solution is to underclock the CPU and/or decrease the voltage a bit until it's stable and throttling no longer occurs. If you have the stock cooler, you could try it to see how much it helps with your issue.

The CPU throttles, but does it shut down when stress testing it for as long or longer than normally is required for the system to power off? If not, then stress test both CPU and the GPU at the same time; if it shuts down, then you probably also have a PSU issue.

 

[maxwellmelon]

Your cpu is overheating.

have you tried reseating the heat sink. redoing the thermal paste????
did you tighten the two screws down (step 4 of your heat sink install guide)???
did you remove the plastic covering the thermal paste if it comes with thermal paste on it during install?


what is your room temp?
what is the case temp?
what is the cpu temp at idle?
what is the cpu voltage at idle?
(fyi the asus motherboard program is good at reading temp values and voltage values)

Your PSU has enough power. I have same PSU, cpu, and a gpu that uses more power. and my unit doesn't get anywhere near the limit.

 

[GhislainG]

I'm not convinced the CPU really overheats (I have yet to see a picture of AOD); it most likely is the VRM that does.

 

[maxwellmelon]

I'm not convinced the CPU really overheats (I have yet to see a picture of AOD); it most likely is the VRM that does.

"During the stress test the Thermal Margin dropped to -11.6 degrees celcius, and the cores often throttled down from 4018 MHz to 1406 MHz, from 20X multiplier to 7X multiplier and from 1.3375 V to 0.8875 V"

the cpu is in neg thermal margin. the cpu is overtemp itself and therefore is throttling. a normal acceptable max load thermal marg would be positive 10c, neg 10c means it has crossed max acceptable die temp and is engaging safety settings.

 

[GhislainG]

I'm not convinced the CPU really overheats (I have yet to see a picture of AOD); it most likely is the VRM that does.

"During the stress test the Thermal Margin dropped to -11.6 degrees celcius, and the cores often throttled down from 4018 MHz to 1406 MHz, from 20X multiplier to 7X multiplier and from 1.3375 V to 0.8875 V"

the cpu is in neg thermal margin. the cpu is overtemp itself and therefore is throttling. a normal acceptable max load thermal marg would be positive 10c, neg 10c means it has crossed max acceptable die temp and is engaging safety settings.

Closely monitoring AOD is very important. A sudden thermal margin drop is caused by the VRM; if the drop is more gradual then it usually is the CPU. The motherboard has a non-heatsinked 4+1 phase VRM and the cooler doesn't blow air on it; therefore it should normally overheat and cause throttling.

 

[r0uge]

I've cleaned out my computer with a hoover (there was a lot of dust inside) and the random shutdowns have stopped. I'm still concerned about my CPU heat/throttling though. I'm getting two different answers here so not really sure what to do

 

[GhislainG]

Why not install a small fan to cool the VRM and see if it stops the throttling issue?