Question GTX 980ti turning off display & crashes PC when on load

[dr34mf13ld]

Hi guys.
This card EVGA GTX 980ti GAMING SC2.0
Bought in 2016 with warranty ended on 26.02.2019
Used for 4K gamin since 2017
Problem first occurred on Saturday 25.01.20 while playing NFS Rivals on highest possible settings on two 4K displays (cloned).

PC specs (or look on the signature):
i7-3770k (not OC)
EVGA GTX 980ti GAMING SC2.0 (not OC)
4x4GB @2133 with XMP Profile
PSU SS-850AM
Display Dell P2715Q / 4K Philips TV

While gaming I'm getting this message on my connected display(s):
"The current input timing is not supported by the monitor display. Please change your input timing to 3840x2160@60 hz or any other monitor listed timing as per the monitor specifications."

The problem doesn't occur during boot, login or when I'm using the PC browsing or watching UHD movies, I tried. It can stay idle and nothing will happen for hours.
I'm using Diablo 3 /which I've spent a thousands hours on the highest settings never having a problem/ as stability test.
The GPU would turn off and crash even on a low resolution of 1920x1080 with Anti-Aliasing turned on (or off with a longer delay). It managed to crash even when I stayed in the menu for longer.

Initially I had "Display driver nvlddmkm stopped responding and has successfully recovered. " critical warning in Event Viewer. Installing the latest NVIDIA drivers fixed that and I'm not getting this message again.

Now the Event Viewer has no critical system messages at all that can point out what exactly is happening and if the GPU is the main cause or just the victim.
*Sometimes I'm getting "Disk 2 has been removed" due to duplicated IDs message, disk2 in this case is my OS drive, but I don't think it will cause the video to stop.

When the display turns off and there's no picture the entire PC freezes, become irresponsible and sometimes even the reset/power buttons on the case doesn't react, something that tells me the problem may be related to the PSU instead.

CPU/GPU temps seems OK. In fact the GPU doesn't even reach higher usage or temps at the time it stops.
I tested the fans with MSI Afterburner etc, they are working when the card's temp go up.

I ran a Furmark stress test on GL 4.5 with 99,9% GPU usage and 78 degrees temp and it passed.
I ran a Tessmark x8 and it crashed in seconds (on Saturday). The test I did on Wednesday caused the GPU to produce strange noises such as blowing (whistling) or when a milk i about to boil in those special pots that start to whistle. The noise is not coming from the fans but the card itself, probably from some "chip" that is being fried..

I tested the card on Windows 10 with the latest nvidia drivers (my current config) - it crashed.
I tested it on another drive Windows 8 with nvidia drivers 381. or older - it crashed.
I tested removing most of the devices, peripherals, external drives etc. to exclude them as possible cause - the GPU crashed.
Under-clocked the GPU - it crashed.
I did a test with different display and DP/HDMI cable - it crashed.
I tried my integrated HD4000 with Diablo 3 that is hardly playable, at 100% usage for an hour there was no crash whatsoever.

Tomorrow I'm going to test it on another system, because I can't identify the problem in mine.

Any ideas? Tips on how to test where's the problem?

Any help will be highly appreciated.

Edit: I've updated a bit the description and title.

 

[helpstar]

Remove the GPU and try stress testing with the onboard graphics again.

Tbh, get the GPU to a different system or a different GPU (similar to the 980ti) for your system.

Did you have a look at the CPU´s temperature as well?

 

[dr34mf13ld]

I'm not at home right now, but tomorrow evening I've got a long list of things to do.

To start with testing the GPU on another OS (I have a forgotten SSD with Win 8 with old drivers and stuff). I'm only using it when my main ssd/Windows 10 cannot boot, but i think it will do the thing to exclude recent software changes made on the other drive.

Additionally I'll remove every other HDD/device and unwanted peripherals that may cause the bug. Will boot with only ssd/gpu/cpu + a stick of ram and see what will happen.

I've talked to a friend about getting my GPU into his system to see how it goes with another PSU.

"Remove the GPU and try stress testing with the on-board graphics again. " - this also. Thanks for the idea.

About the CPU - I haven't checked, but if I run a stress test on it and it doesn't fail in 5 seconds (up to 5 minutes) it's not it.

I found a similar thread to mine, spoiler: The card smoked and died

 

[stonecarver]

I

Dude, I'm having the same issue.

Is your Card EVGA ?

Sorry I didn't see this thread.

With the card out with good light look at the back of the board a lot of failing EVGA card have a copper heat strip that almost looks like copper tape soldered on to pull heat off the two devices it's soldered to.

As time go on and heat from using card that connection fails and the copper tape just kinda burns away.

If you have this issue it needs to be fixed Before the card fails.

 

[oldcracc]

It sounds like your GPU is probably dead, other than that it could be a PSU issue, I wouldn't expect it due to your system working completely fine with the same PSU, but I wouldn't throw it out of the equation.

 

[dr34mf13ld]

I did few tests today.
Tested the GPU with my 2nd OS Windows 8 and old Nvidia drivers (381) - it failed.
Tested with all unnecessary hardware removed - it failed.
Tested with HDMI cable instead of DP - it failed.
Tested underclocking the card - it failed.
Removed the card, cleaned the dust, cleaned the PCI-E slot - failed.

Tested the integrated HD4000 with the same conditions on a 100% usage - there was no crash.

I definitely have to test the card on another PC because I'm stuck with mine.
If the problem was a defected MOSFET or copper strip the card would be dead by now.
Something that I noticed today was when running Tessmark x8 test on the geeks3d gpu test there was this noice coming from the GPU like someone is blowing, but it wasn't coming from the fan but the card itself. No need to say I immediately stopped the test.
For the record I did Furmark test twice today and on Sunday on Open GL 4.5 with 99,9% usage and 78 degrees on the card and it passed twice. But every game would crash it even if I stay in the menu for like 3 minutes.

I can still use most of the time with no gaming, and watching UHD movies obviously doesn't break it, but still no idea what's going on.