Artifacts

[Johnny999]

I've noticed artifacts on the screen today (GPU is RTX 3070 Gaming X Trio, not overclocked whatsoever) - black/green boxes, weird pixels, black vertical lines etc. It went away after I rebooted the PC. Currently I am running the OCCT VRAM test for 30 minutes (still 30 minutes remaining) and so far it has not detected any errors. I have checked Event Viewer and there are 3 events (located between the boot and reboot) with the source being Display. It's written there that Display driver nvlddmkm stopped responding and has successfully recovered. There are only 3 entries with this error in Event Viewer since I installed Windows 10 back in August last year, all of them today. Could it be a driver issue? I have the latest 461.40 driver installed on 26th January this year.

[Guest]

If you're lucky it's just a corrupted driver. Grab the newest driver and use DDU, rebooting after every step.

https://www.guru3d.com/files-details/display-driver-uninstaller-download.html (mirrors at the bottom)

https://www.nvidia.com/Download/driverResults.aspx/170312/en-us

 

Edit: I skipped the last line lmao, still download a fresh version

[Johnny999]

29 minutes ago, Sp4gh3tti said:

If you're lucky it's just a corrupted driver. Grab the newest driver and use DDU, rebooting after every step.

https://www.guru3d.com/files-details/display-driver-uninstaller-download.html (mirrors at the bottom)

https://www.nvidia.com/Download/driverResults.aspx/170312/en-us

 

Edit: I skipped the last line lmao, still download a fresh version

OCCT found no errors (182 passes). I'm going to try DDU.

[Guest]

5 minutes ago, Johnny999 said:

OCCT found no errors (182 passes). I'm going to try DDU.

That bodes well for the physical health of the card. Again make sure you reboot after DDU (it gives you the option to not)

[Johnny999]

1 minute ago, Sp4gh3tti said:

That bodes well for the physical health of the card. Again make sure you reboot after DDU (it gives you the option to not)

The program recommended that I don't select 100% VRAM for testing because of other programs. It was like 6400 MB or something. Did it test all 8192 MB over the passes?

[Guest]

1 minute ago, Johnny999 said:

The program recommended that I don't select 100% VRAM for testing because of other programs. It was like 6400 MB or something. Did it test all 8192 MB over the passes?

Probably not (guessing, not sure how it works), but whatever it couldn't touch was likely already in use by Windows and would likely be the first used when loading a game

[Guest]

This may be silly but you tried reseating it? Could've wiggled loose over time

[Johnny999]

1 minute ago, Sp4gh3tti said:

Probably not (guessing, not sure how it works), but whatever it couldn't touch was likely already in use by Windows and would likely be the first used when loading a game

I've sent them (OCCT) a mail, hopefully they will reply.

[Guest]

The safe play here is immediately RMA the card

 

Have you tried slamming it with a heavy synthetic 3D load like Furmark or maxed out Superposition? If you can trigger artifacts again RMA the card

Silicon lottery exists, none of us are safe

[Johnny999]

9 minutes ago, Sp4gh3tti said:

The safe play here is immediately RMA the card

 

Have you tried slamming it with a heavy synthetic 3D load like Furmark or maxed out Superposition? If you can trigger artifacts again RMA the card

Silicon lottery exists, none of us are safe

Shouldn't I wait first and see if DDU worked? I don't have any other graphics card and my CPU is not an APU.

[Johnny999]

10 minutes ago, Sp4gh3tti said:

The safe play here is immediately RMA the card

 

Have you tried slamming it with a heavy synthetic 3D load like Furmark or maxed out Superposition? If you can trigger artifacts again RMA the card

Silicon lottery exists, none of us are safe

How long should I let Furmark run?

[Guest]

DDU is just a utility for making sure every piece of the driver is removed since they love to leave shit behind. Unless you mean before RMAing in which I don't really think it matters as long as you're within warranty. After DDU and the driver reinstall hit it with Furmark (or similar) for a while and see if you can trigger the artifacts again.

I mean like, a while. You'll want to see it achieve 60c easily and maintain max boost clocks at around that temp.

 

If it detected a hardware failure it may have locked the card to a lower power state where the artifacting doesn't happen, making it seem like a one time thing. That's why watching the clocks under load is crucial

[Guest]

Base Clock   (at the very least hitting this)

1500 MHz

Boost Clock   (should still get this around 60c)

1830 MHz

Memory Clock  (may be slightly lower but it should peak here)

1750 MHz

 

 

Edit: These may be slightly inaccurate but whatever

[Johnny999]

40 minutes ago, Sp4gh3tti said:

Base Clock   (at the very least hitting this)

1500 MHz

Boost Clock   (should still get this around 60c)

1830 MHz

Memory Clock  (may be slightly lower but it should peak here)

1750 MHz

 

 

Edit: These may be slightly inaccurate but whatever

It reached about 1755, here are the results: https://gpuscore.top/furmark/show.php?id=288318

There were no artifacts.

[Guest]

16 minutes ago, Johnny999 said:

It reached about 1755, here are the results: https://gpuscore.top/furmark/show.php?id=288318

There were no artifacts.

How long did you let it run? Your score is right on par with other 3070s so it's probably hitting max boost

idk friend, the safe move here is just RMA it at the first sign of trouble but it's rather strange it happened once and can't be recreated. If it happens again I'd cut your losses and take the free replacement.

[Johnny999]

2 minutes ago, Sp4gh3tti said:

How long did you let it run? Your score is right on par with other 3070s so it's probably hitting max boost

idk friend, the safe move here is just RMA it at the first sign of trouble but it's rather strange it happened once and can't be recreated. If it happens again I'd cut your losses and take the free replacement.

I used the 1080p preset which lasted for a minute.

[Guest]

Just now, Johnny999 said:

I used the 1080p preset which lasted for a minute.

Try superposition, run it like a dozen times. Idk if Furmark is still being updated to be as demanding as it once was

[Johnny999]

51 minutes ago, Sp4gh3tti said:

Try superposition, run it like a dozen times. Idk if Furmark is still being updated to be as demanding as it once was

Well that was quite a bit more intense than Furmark IMO. I haven't noticed any artifacts and there aren't any Display logs in the Event Viewer.

[Guest]

Strange indeed. Again, safe play is RMA, I have to keep repeating that since artifacting is usually a sign of impending card death and not immediately suggesting RMA could be taken as "bad advice". But imo the card would've shit the bed again after stressing like this, the fact that it didn't leads me to believe the one time it happened could've just been a fluke? I forget if I already asked but you tried reseating the card I assume?

[Johnny999]

4 minutes ago, Sp4gh3tti said:

Strange indeed. Again, safe play is RMA, I have to keep repeating that since artifacting is usually a sign of impending card death and not immediately suggesting RMA could be taken as "bad advice". But imo the card would've shit the bed again after stressing like this, the fact that it didn't leads me to believe the one time it happened could've just been a fluke? I forget if I already asked but you tried reseating the card I assume?

Yeah I did, it's a pain doing it though because it's big and there's also the anti-sag metal bracket under it. I'll wait and see if it happens again.

[Guest]

44 minutes ago, Johnny999 said:

Yeah I did, it's a pain doing it though because it's big and there's also the anti-sag metal bracket under it. I'll wait and see if it happens again.

Even if you continue forward with absolutely zero issues I'd RMA it in the final days of your warranty, just to be sure

Perhaps consider a permanent -5% power target?

[Johnny999]

15 minutes ago, Sp4gh3tti said:

Even if you continue forward with absolutely zero issues I'd RMA it in the final days of your warranty, just to be sure

Perhaps consider a permanent -5% power target?

I've never fiddled with Afterburner, I'll just leave it as it is and see what happens. Thanks for your advice.

[Guest]

Happy to (try to) help!

[Johnny999]

20 hours ago, Sp4gh3tti said:

Happy to (try to) help!

No artifacts today, and I've done some more OCCT testing (about 80 minutes, 90-95% VRAM) - no errors. We will see.

[papajo]

 

1 hour ago, Johnny999 said:

No artifacts today, and I've done some more OCCT testing (about 80 minutes, 90-95% VRAM) - no errors. We will see.

 

Do you use a water cooler for your CPU ? if that is the case check for leakages(get a piece of toilet paper gently touch along the surface and see if it absorbed any moisture do the same for the pipes )  maybe a drop fell down on your GPU and now it vaporized and the GPU works ok again.