Question Display loss of signal/black screen, GPU fans 100% in older games only

[nitrocide]

A strange problem I've had for months and have not found a fix whenever I search around. It seems a lot of people have had this happen but I haven't found anyone with my unique situation. After being in game for usually 20 minutes or so, sometimes maybe a little longer, I get a sudden loss of monitor signal (screen goes black, light turns orange) and my GPU fans go to a loud constant 100% speed. Ctrl+Shift+Win+B does nothing after this happens (restart video driver). This has only ever happened in older games, specifically Team Fortress 2 and Far Cry 2. I have played GTA V, Valheim, Mad Max, Horizon Zero Dawn and more for hours without ever having the issue. I watched temps and voltages with HWMonitor while playing Far Cry 2 and they stay completely fine, but I also don't know how these could be the issue since newer games don't cause the problem.

Things I've tried that didn't fix it:
I tried older drivers but not extremely old because I think it would cause worse performance on my already sub-par GPU in the newer games that play fine
I opened my GPU and dusted it and replaced the thermal paste

Specs:
Windows 10 64 bit
ROG Strix Z370-E Gaming
i7-8700k @ 4.0GHz
16GB Corsair Vengeance 3000mhz
EVGA GTX 770 2GB SC
620W Antec HCG
1TB WD HDD and 120GB Samsung 840 SSD

 

[Lutfij]

How old is the PSU in your build? Considering that the GPU you own is quite dated, it's possible that the GPU is on it's way out, due to wear and tear(and age) or that your PSU is starting to show it's age as well or that the drivers are not communicating well with the GPU.

In order to rule out the PSU, you will need to source a PSU from a friend or neighbor who owns a reliably built PSU and one that has at least 650W of power at your entire system's disposal. If the donor PSU doesn't bring the problem up, the issue was with your PSU.

In order to rule out your GPU, you should first try and make sure your motherboard is on the latest BIOS update and that you're on the latest OS version, meaning 21H1 for Windows 10. use DDU to uninstall your GPU drivers and then reinstall with the latest driver version, in an elevated command, i.e, Right click installer>Run as Administrator.

You can also take the GPU to your friend/neighbors system to see if the issue recurs on their end as well.

Side note, is C6 state mode in BIOS disabled?

 

[nitrocide]

How old is the PSU in your build? Considering that the GPU you own is quite dated, it's possible that the GPU is on it's way out, due to wear and tear(and age) or that your PSU is starting to show it's age as well or that the drivers are not communicating well with the GPU.

In order to rule out the PSU, you will need to source a PSU from a friend or neighbor who owns a reliably built PSU and one that has at least 650W of power at your entire system's disposal. If the donor PSU doesn't bring the problem up, the issue was with your PSU.

In order to rule out your GPU, you should first try and make sure your motherboard is on the latest BIOS update and that you're on the latest OS version, meaning 21H1 for Windows 10. use DDU to uninstall your GPU drivers and then reinstall with the latest driver version, in an elevated command, i.e, Right click installer>Run as Administrator.

You can also take the GPU to your friend/neighbors system to see if the issue recurs on their end as well.

Side note, is C6 state mode in BIOS disabled?

Thank you for replying, I'm working on getting windows updated from 20h2 to 21h1, and then will flash the latest bios from March. I had an issue with my pc shutting down while playing games about 2 years ago and I updated the bios and never had the issue again, so I'm hoping that will be the fix for this too. Seems unlikely to be failing hardware since it only happens in less intensive games, but my PSU is from 2012. I checked for C6 state in bios and didn't find that specifically but found a setting for which cores to enable and it is set to "all cores enabled".

 

[nitrocide]

Windows update/bios update didn't fix it. Seems the problem was Asus AI Suite, uninstalled it and was able to play for much longer with no problem. My guess is it causes some conflict with fan speed/temperature monitoring.

 

[nitrocide]

So that didn't end up being the fix, and I have still been trying to figure this out, got a new psu and still have the issue but I just discovered something. I have HWMonitor running to watch my gpu temperature while I start up a game. If I push up on the end of my gpu where it sags a little, the temperature reading goes to ~90C in HWMonitor. When I take my hand off and let it sag again, the temperature goes to ~60C. So it is actually overheating but the temperature reading is wrong while the gpu is sagging. I'm thinking I should be able to fix this by opening my gpu, but I don't really know what to look for.

 

[dotas1]

So that didn't end up being the fix, and I have still been trying to figure this out, got a new psu and still have the issue but I just discovered something. I have HWMonitor running to watch my gpu temperature while I start up a game. If I push up on the end of my gpu where it sags a little, the temperature reading goes to ~90C in HWMonitor. When I take my hand off and let it sag again, the temperature goes to ~60C. So it is actually overheating but the temperature reading is wrong while the gpu is sagging. I'm thinking I should be able to fix this by opening my gpu, but I don't really know what to look for.

Your assumption is false. You lifted the GPU from a side that probably made the cooler have less contact with the die or the VRAM. A gpu can heat up that much in less than a second when you move the cooler and makes improper contact.
If you try lifting the GPU from a different part or from the middle that the cooler gets somewhat even pressure, you will either not have a temp difference, or a much smaller one.

 

[nitrocide]

Your assumption is false. You lifted the GPU from a side that probably made the cooler have less contact with the die or the VRAM. A gpu can heat up that much in less than a second when you move the cooler and makes improper contact.
If you try lifting the GPU from a different part or from the middle that the cooler gets somewhat even pressure, you will either not have a temp difference, or a much smaller one.

I tried it, same temperature increase when lifting it from the left side, middle, or evenly on both sides. There's a game I can load and cause the loss of signal/100% fans every time. What's weird is I don't get the loss of signal/fans 100% when lifting the gpu but I do otherwise.