Question Took apart my GPU to clean/repaste. Games now crashing to desktop.

[tcoster21]

So initally when I reinstalled the GPU, I got stuck in a boot loop. I took everything apart, cleaned it, and put it back in and it started up fine. Then a few hours into gaming I started getting a drop in fps to 10 and then crash to desktop. It would eventually close but my fps would remain at 10 unless i killed the task manually. I'm not sure at this point if it's a GPU issue or a MoBo issue. I've checked the gpu many times and don't see any issues. Everything runs fine, no temp spikes or power spikes/drops, I cannot figure this out! Only thing I haven't tried is reflashing the bios of the GPU as well as the MoBo.

 

[Lutfij]

Welcome to the forums, newcomer!

Considering that this is a thread of troubleshooting nature, you're advised to state your full system's specs. Include your PSU's make and model and it's age. By clean, did you make sure to only remove the thermal paste and not the pads? I'm assuming the worst here since there's no mention of the discrete GPU's make and model in your build that you'd torn down.

I would advise not to flash the VBIOS. Instead see if you can download the latest drivers for your card from manufacturer's support site and then use DDU to uninstall your GPU drivers. Reinstall the latest drivers in an elevated command, i.e, Right click installer>Run as Administrator, and see if that helps.

 

[tcoster21]

Apologies. I have posted so many times with this issue I've left out a lot of info. I have tried DDU. I have tried 4 different drivers. I have repasted again, I have not removed any thermal pads (although they might have shifted a teensy bit) but I have monitored the thermals during use and they do not go over 50. I have tried different power cables (modular psu). I have removed my OC on my cpu and my ram. I have tried increasing the power limit of the gpu through X1. Everything was working fine until I tried to put it back in after taking it apart. I've never had a problem like this before. I can leave my PC all day and it's fine, but as soon as I game - either 20 min or 3 hours, it will always crash to desktop at 10 fps. I have unplugged all extra usbs and kept it as simple as possible.

Also - I do have a discrete GPU but I've never been able to get it to work. I've always ignored it and never had issues before repasting. I'm thinking something shorted causing the boot loop when I initially reseated.

When using DDU, I reboot into safe mode, stop windows from downloading updates, restart and install the driver.

PC is 16 months old. PSU is 8 months old. GPU is 1 year old.

Specs:
i9-9900k
EVGA 2080 Super FTW3
2x 16gb Corsair Vengeance RGB 3200 DDR4
NVME Samsung 500gb SSD (OS installed here)
SATA SSD Samsung Evo 860 1tb
Corsair 850gold Rmx something or other.
windows 10 pro
Corsair h115i Platinum 280mm AIO

Peripherals:
Razer Hunstmans Elite
Razer Nari Ultimate Headset
Logitech G502 mouse
Razer RGB Mousepad
Logitech G923 Wheel and Pedals

 

[tecmo34]

I would look at replacing the thermal pads and thermal paste on the GPU (since you are comfortable doing so) and make sure everything is back together correctly. I think it is always wise to replace thermal pads when doing a tear down such as you did for cleaning. The temps might not look bad but doesn't mean components aren't overheating that the sensors aren't monitoring.

 

[tcoster21]

I would look at replacing the thermal pads and thermal paste on the GPU (since you are comfortable doing so) and make sure everything is back together correctly. I think it is always wise to replace thermal pads when doing a tear down such as you did for cleaning. The temps might not look bad but doesn't mean components aren't overheating that the sensors aren't monitoring.

So I have done this 3 times now. SOOOOO tired of removing 100 screws from the backplate lol. The pads all look good, I don't have access to replacements but I can look into it. I also checked the temps off each individual temp sensor in the card, there's 6 I believe, and not one spiked or showed any outliers. I don't think it's a temp issue, in all honesty. I flip flop between thinking its the gpu and something else. I don't know why it would fail only in games with nothing to show me what might be going wrong!

EDIT: In regards to "teardown" - I literally took off the back plate, removed the 3 cables attached to the board, removed the paste, reapplied, reattached the backplate to the board. That's it.

 

[Vic 40]

What motherboard? Is there a second pcie slot you can use to test?
Tried with two pcie power cables? =not one/two connectors, or maybe already using that ...

 

[tcoster21]

What motherboard? Is there a second pcie slot you can use to test?
Tried with two pcie power cables? =not one/two connectors, or maybe already using that ...

ASUS ROG Strix Z390-E is the MOBO. I've tried both PCI-E slots. I also tried using my CableMod vertical mount with the riser cable just for kicks. No joy there. I had 2 separate power cables initially. I tried using just one daisy chained and then the other and it still failed. I haven't tried a new psu (I have my old 650 still). But I don't think that would be the problem, but I can try that PSU hooked up directly to the GPU and see if it fails. I can also try a different cable too but I don't know why both of them would be bad that I tried. I will say this - I have run many stress tests and only error I ever saw on the GPU side of things was in OCCT and after 3 hours it showed a VRAM error. I couldn't get it to duplicate again after that time. My games also crash usually before an hour goes by, if that means anything. However, before I sold my 2080ti I stress tested it and got the same errors from it on one pass I coudln't duplicate so IDK may be a pointless comment there lol.

 

[tcoster21]

Also - I don't think this matters but there is a lot of paste still stuck around the GPU chip. How do i explain this, hmm...so like the little gold metal "bumps" that line the chip, have paste caked around them. I don't know if this matters. I tried to clean it but it's in there pretty good.

 

[Danra]

Excess thermal paste usually causes no problems, unless the paste is electrically conductive. I say usually because on one occasion in my 35 years of working in the computer field I did find a computer that had massive amounts of TIM globbed all around the CPU on the motherboard and it caused the computer to not work until I cleaned up the mess. To remove old TIM apply ArctiClean 1 and let it set a few minutes then clean off the mess with a cloth. If the caked on TIM is still stuck in spots you can use a cheap clean tooth brush with ArctiClean 1, then wipe the surface with a cloth.

Next, apply ArctiClean 2 and clean it off with a cloth, reapply ArctiClean 2 and remove with a coffee filter to remove any cloth or fiber left behind. If you want it squeeky clean after that you can use 91% alcohol and rub it over the surface with a coffee filter after you are finished with ArctiClean 1 & 2. You can get ArctiClean in various places, newegg, amazon and so on. I have used ArctiClean for decades and it has always done a great cleaning job for me.

And, only use cables that came with the power supply. Some of them have different wire thicknesses and some have inline capacitors.

 

[tcoster21]

Excess thermal paste usually causes no problems, unless the paste is electrically conductive. I say usually because on one occasion in my 35 years of working in the computer field I did find a computer that had massive amounts of TIM globbed all around the CPU on the motherboard and it caused the computer to not work until I cleaned up the mess. To remove old TIM apply ArctiClean 1 and let it set a few minutes then clean off the mess with a cloth. If the caked on TIM is still stuck in spots you can use a cheap clean tooth brush with ArctiClean 1, then wipe the surface with a cloth.

Next, apply ArctiClean 2 and clean it off with a cloth, reapply ArctiClean 2 and remove with a coffee filter to remove any cloth or fiber left behind. If you want it squeeky clean after that you can use 91% alcohol and rub it over the surface with a coffee filter after you are finished with ArctiClean 1 & 2. You can get ArctiClean in various places, newegg, amazon and so on. I have used ArctiClean for decades and it has always done a great cleaning job for me.

And, only use cables that came with the power supply. Some of them have different wire thicknesses and some have inline capacitors.

I usually use alcohol and coffee filter for paste. But this stuff is caked on there so it wasn't doing much, toothbrush is a good idea. I've tried 2 different cables so far on the GPU. The paste is Kryonaut which is non-conductive so I didn't think it would be causing a problem.

 

[Ash95]

Is it only the one game?

Also, would be handy if you shared benchmark results.

i’ve had a similar issue before and i had to uninstall steam and redownload the game and it worked. Although my games was a strategy game i believe.

 

[Kona45primo]

Check Nvidia Control Panel -> Configure Surround, PhysX -> make sure PhysX settings Processor: is "Auto-select"

 

[tcoster21]

Is it only the one game?

Also, would be handy if you shared benchmark results.

i’ve had a similar issue before and i had to uninstall steam and redownload the game and it worked. Although my games was a strategy game i believe.

No it's all games. F1 2020 used to run flawlessly, now it crashes as well.

What would help to post? I'll put up whatever is necessary.

 

[tcoster21]

Check Nvidia Control Panel -> Configure Surround, PhysX -> make sure PhysX settings Processor: is "Auto-select"

I'll double check that when I get home, I think it's set to that but I will look.

 

[tcoster21]

OK - Is there a program I can run that can log what happens when the game crashes? I do get an error log from F1. I cannot duplicate the problem in ANY testing software. I've run 4 different tests and not one gives me an error. I'm really at my wits end, I'm about to bring it to a shop and see if they can figure it out. The only thing I can think is there might be a memory leak as once the game crashes and even if I kill it, my fps still stays low. Then a few minutes later itll recover. I'm attempting to adjust the paging file size and trying an even older driver to see what happens.

 

[tcoster21]

increased the paging size let me play for about 30 minutes before it crashed. this time i had audio on so when it crashed the error log came up and i coudl still hear everything in the background. This time after submitting the error log my FPS did go back to normal right away.

 

[tcoster21]

UPDATE:

Had my card RMA'd - problem still persists. I think something happened in the latest windows update as everything was fine with the new card until I had to update so I could install the drivers for the xbox wireless adapter. I played yesterday evening for about 3 hours or so without a crash. 2 different games. Before that they were crashing every hour or so. Straight to desktop, no error, no temp spikes, nothing. So I set my bios settings back to their defaults and it seems to be working so far. I don't know what would cause my OCs to all the sudden cause crashes. Maybe some update or new driver did something screwy. Oh, I also deactivated ALL windows "game" things like the game bar, their new "graphics settings" options, and also their "game mode". So, maybe it's that that has seemed to fix it....so far at least. Well, just wanted to put an update in here.