GPU Issue

Hector45 · 09 Apr 2022

Hello all, I've been trying to track down the cause of a CTD issue I have been having seemingly out of now where in all games. I suspect there is an issue with the GPU at this point. Usually I will get a CTD after playing about 1-2 hours. What I've tried so far:

Removed drivers with DDU tool and clean install.
Tried several different drivers.
Updated firmware
Removed all CPU overclock
Applied 60FPS screen cap
Reduced GPU power target
Downclocked GPU by 40Mhz
Removed and re-seated GPU

Recently I've tried using EVGA's PX1 stress test tool and when I get about halfway done with the test I get a hard crash and restart (dump file attached). GPU is an EVGA 3080Ti FTW3, PSU is new EVGA 850watt, All 3 GPU power cables are separate from the PSU (no daisy chain).
Anyone have any idea what it could be, or do you think it is the GPU?

Steve C · 09 Apr 2022

Could a faulty GPU. Can you try it on a different PC?

bobkn · 09 Apr 2022

Sorry, I'm not much at reading dump files.

Have you tried using GPU-Z Graphics Card GPU Information Utility to monitor the GPU temperatures while you're stress testing it?

I'm not optimistic that will help, as the card ought to throttle when it overheats.

I see that the eVGA nVidia RTX 3080Ti FTW 3 Ultra has a factory overclock. You could try downclocking it a little (Access Denied) to see whether that changes anything. I've had lockups when I have OC'd my GPU (regular RTX 3080). It'd be disappointing if your 3080 ti couldn't meet eVGA's specs.

Pejole2165 · 09 Apr 2022

Try HWInfo64 and set it to monitor voltages for both CPU and GPU, also temperatures, use the logging feature so you can view recorded values after a crash.
See if there are any events (event viewer) related to drivers, power issues, corresponding to crash times.
Eliminate as many variables as possible, so don't have any 3rd party apps running (other than those for logging temps/ voltages).
There is an issue with some RGB apps that provide lighting settings for certain SSDs (M.2 ones mainly, I believe) that include the driver regardless of whether the hardware is present, it is known to cause lockups, bluescreens and CTDs.
Autoruns from Sysinternals (or the MS Store) can help to temporarily disable 3rd party drivers, scheduled tasks etc to help with troubleshooting.

Hector45 · 12 Apr 2022

So I switched GPU's with my brother, he has an EVGA 3080 XC3 and just got a CTD, event log screen cap below, any ideas?

GPU Issue-image_2022-04-12_195807181.png

sportster · 12 Apr 2022

After swapping graphics card i assume that you have DDU and reinstall driver ? From the error log it seem
there is a problem with the game itself. You can try full uninstall and reinstall the game and see if it help.

Hector45 · 12 Apr 2022

Tried a different game, CTD after about an hour. I've tried so many things at this point I don't know what's left besides reinstalling windows.

Steve C · 13 Apr 2022

If the GPU crashes on two PCs then suspect a faulty GPU.

Pejole2165 · 13 Apr 2022

Have you monitored the temps and voltages yet? 30 series cards are known to have power spikes of over 100 watts at times, these spikes can come often enough to cause the power supply to not be able to keep up, the card then doesn't receive the power it expects which in turn can cause all manner of lockups, crashes and strange behaviour.
Since you claim the crashes occur after some time of gaming it would suggest your issue may be temp related rather than power though, how is the airflow in both systems you have tried the card in, have you tried custom fan curves?
Downclocking by such a small amount on such a powerful card is not really going to show anything much.
You can use MSI Afterburner in conjunction with HWInfo64 (they can hook into each other via RTSS) to monitor temps and voltages across the system.
Afterburner can be used to adjust power and temp limits and memory and core clocks in real time while monitoring, maybe also using Unigene's Heaven benchmark to see the effects of changes. Both Afterburner and HWInfo64 can be left running in the background to monitor the system while gaming and may give a clearer reason for the crashing.
It might be a good idea to uninstall PrecisionX first as that installs a driver that may conflict with other, similar, tools.
Jayz2cents has a very good video explaining the process on his YouTube channel if you are interested and uses EVGA 30 series cards for examples.

It is entirely up to you if you want to do some testing for yourself rather than try to RMA in the current climate, but none of the above can physically damage the card as it has built in protections and limits of it's own.

Hector45 · 13 Apr 2022

Steve C said:

If the GPU crashes on two PCs then suspect a faulty GPU.

I'm currently using my brothers 3080 in my machine and still getting crashes. My 3080Ti has been running fine in his machine.

- - - Updated - - -

Ran HWinfo64, RDR2 seems to get it to CTD quickest. Looked through the log but couldn't find anything that stood out unless i missed it.

- - - Updated - - -

Interesting, the last two CTD in the event log i see a Warning "Display driver nvlddmkm stopped responding and has successfully recovered". This happens about 30 seconds before the crash event.