Monitors turn off and GPU fans kick to 100% while gaming

[Firestarya]

Issue started about 3 days ago, seemingly out of nowhere and ONLY occurs when I'm playing a game (did not update or change anything recently.) I have tried 4 different games, all of which result in the same crash after anywhere from 3 mins-10 mins. After the crash occurs, I can still hear audio but am forced to hold the power button to restart my PC. I'm at a loss here and not sure what to do, so any help would be greatly appreciated.

My system specs from Speccy are:

Operating System
Windows 10 Home 64-bit
CPU
AMD Ryzen 7 3700X 56 °C
Matisse 7nm Technology
RAM
32.0GB Dual-Channel Unknown @ 1066MHz (15-15-15-36)
Motherboard
Micro-Star International Co. Ltd. MPG X570 GAMING EDGE WIFI (MS-7C37) (AM4) 43 °C
Graphics
27GL850 (2560x1440@144Hz)
LG IPS FULLHD (1920x1080@60Hz)
4095MB NVIDIA GeForce RTX 2080 SUPER (Gigabyte) 57 °C
Storage
1863GB Seagate ST2000DM008-2FR102 (SATA (SSD)) 33 °C
931GB Seagate ST31000528AS (SATA ) 33 °C
931GB Samsung SSD 970 EVO Plus 1TB (Unknown (SSD))
953GB Sabrent (Unknown (SSD))
Optical Drives
No optical disk drives detected
Audio
NVIDIA Virtual Audio Device (Wave Extensible) (WDM)

Things I have tried:

• Using DDU to uninstall NVIDIA GPU drivers and reinstalling​
• Rolling back to a previous GPU driver version​
• Replacing the 6-pin and 8-pin power cable connecting the PSU and GPU​
• Moving the position of the power cable on the PSU to another port​
• Re-seating the GPU and blowing dust out​
• Updating BIOS​
• Reseting BIOS to default settings​
• Reinstalling Windows​
• Purchased thermal paste to replace the factory thermal paste on the GPU (Should arrive 3 days from now)​

Attached is a log from GPU-Z that ends at the time of crash

 

[Bill_Bright]

I recommend you leave the GPU's TIM (thermal interface material) alone. 57°C is barely warm so clearly GPU heat is not an issue. Needlessly changing out TIM needlessly risks damage to the processor due to accident or abuse.

If you still suspect heat, open the side panel of your computer and blast a desk fan in there.

What I don't see in your system specs is anything about your power supply. I always want to ensure I am supplying good, clean stable power before I start messing around with other hardware. I recommend you swap in another, known good PSU to see what happens.

 

[EarthDog]

2133 RAM.. yikes... leaving performance on the table there, bud.

As far as your issue... almost sounds like a bad card. When you list 57C... is that at idle? What are load temps? I don't see a point in voiding your warranty to apply new paste.

RMA the card if possible is my take.

 

[ThrashZone]

Hi,
What is and how old is the psu ?
Sounds like the most likely candidate to me or wall outlet see what is also connected to this circuit and what is the breaker 10-15-20amp.

 

[Vya Domus]

Sounds like power issues but it could be something else just as well. My PC behaved in the exact same manner when I was testing RAM OCs, monitor would lose signal and fans would ramp up.

 

[iuliug]

Do you have OC on CPU or GPU?

 

[TheLostSwede]

A 4GB 2080 Super?

 

[Assimilator]

A 4GB 2080 Super?

I'm guessing OP is using some ancient tool that doesn't know that GPUs can have more than 2^32 bytes of VRAM...

 

[ThrashZone]

Hi,
Typo get over it.

 

[londiste]

Sound continuing and GPU fans ramping up clearly points at GPU being the culprit.

Try Win+Ctrl+Shift+B when it happens.
This is the keyboard shortcut to restart graphics drivers.

 

[Firestarya]

Hi,
What is and how old is the psu ?
Sounds like the most likely candidate to me or wall outlet see what is also connected to this circuit and what is the breaker 10-15-20amp.

My PSU is a Corsair RM750x, same age as the GPU, 11 months. My RAM is at 3600MHz, it was running at 2133MHz because it turned off XMP when I reset my BIOS. It is now back to 3600MHz.

A 4GB 2080 Super?

It is 8GB, seems to be a software error I overlooked when copy pasting lol

 

[TheLostSwede]

Do be aware that if you remove the cooler on your graphics card (to replace the thermal paste) you'll void the warranty.

 

[ThrashZone]

My PSU is a Corsair RM750x, same age as the GPU, 11 months. My RAM is at 3600MHz, it was running at 2133MHz because it turned off XMP when I reset my BIOS. It is now back to 3600MHz.


It is 8GB, seems to be a software error I overlooked when copy pasting lol

Hi,
Memory can do a lot of weird stuff
Maybe test with it with bios optimized default for a little while
Also use sfc /scannow see if it fixes anything.

I'm not a fan of DDU personally.

 

[Selaya]

Try lowering the Core Clock with MSI Afterburner or something

 

[Caring1]

I recommend you leave the GPU's TIM (thermal interface material) alone. 57°C is barely warm so clearly GPU heat is not an issue.

57C might be the reported temp for the GPU Core, and not the VRM section which can be quite a bit warmer and cause the reported issue.
A thermal imaging gun (FLIR) could tell if there are any hotspots that are excessively warm.

 

[Firestarya]

Hi,
Memory can do a lot of weird stuff
Maybe test with it with bios optimized default for a little while
Also use sfc /scannow see if it fixes anything.

I'm not a fan of DDU personally.

I’ve ran Memtest86 with XMP off and on and both came back with 0 errors. And I’ve tried the SFC and DISM commands, nothing came up.

 

[joemama]

what's the temperature of GPU during gaming?

 

[Selaya]

That's not a RAM issue; not even necessarily a GPU temp issue - looks like its a factory OC'd GPU and it just clocked itself to unstable clocks and crashed.

In the Event Viewer there should be an event:

Display driver nvlddmkm stopped responding and has successfully recovered.

There's also these unusually high core clocks right before it craters

2020-09-07 18:38:43 , 2025.0 , 1937.7 , 57.0 , 39 , 1207 , 3083 , 27 , 20 , 0 , 0 , 158.6 , 143.6 , 0.0 , 0.0 , 35.8 , 11.8 , 40.4 , 12.0 , 82.1 , 12.0 , 0.0 , 63.4 , 16 , 1.0500 , 56.1 , 5887 ,
2020-09-07 18:38:44 , 2025.0 , 1937.7 , 59.0 , 39 , 1221 , 3318 , 58 , 28 , 0 , 4 , 178.2 , 162.4 , 0.0 , 0.0 , 40.6 , 11.7 , 45.1 , 12.0 , 92.2 , 12.0 , 0.0 , 71.3 , 16 , 1.0500 , 55.5 , 5908 ,
2020-09-07 18:38:49 , - , - , 0.0 , - , - , 3318 , - , - , - , - , - , - , - , - , - , - , - , - , - , - , - , 0.0 , 0 , 0.0000 , 51.0 , 5903 ,
2020-09-07 18:38:50 , - , - , 0.0 , - , - , 3157 , - , - , - , - , - , - , - , - , - , - , - , - , - , - , - , 0.0 , 0 , 0.0000 , 60.1 , 5968 ,

So yeah, back to my initial suggestion:

Try lowering the Core Clock with MSI Afterburner or something

 

[Toothless]

Hi,
Typo get over it.

Hi,
It's been known that Speccy has not been updated for anything recent for years now. It's not a typo.

 

[Firestarya]

That's not a RAM issue; not even necessarily a GPU temp issue - looks like its a factory OC'd GPU and it just clocked itself to unstable clocks and crashed.

In the Event Viewer there should be an event:There's also these unusually high core clocks right before it cratersSo yeah, back to my initial suggestion:

I did see that in Event viewer! Any idea what I should set my core clock to?

 

[Caring1]

This sounds like a driver issue that has been affecting some cards.

 

[Selaya]

I did see that in Event viewer! Any idea what I should set my core clock to?

Afterburner -> CTRL-F -> CTRL-drag the curve to lower overall clocks a bit, test for stability, rinse and repeat
(or you could just adjust it otherwise, forgot how that worked, I have a custom curve set)

 

[Assimilator]

If you have to underclock your GPU below its stock clocks to get it to be stable, it's defective and should be RMA'd.

 

[Selaya]

2,025MHz is definitively way above stock for the 2080S.

 

[Firestarya]

2,025MHz is definitively way above stock for the 2080S.

It’s at 1650MHz according to MSI Afterburner, lowered it by -165 and it lasted a bit longer, but still black screened after about 10-15 minutes. I then put a curve on that lowered it by -200 and it black screened at the normal speed.