RTX 3000 Series artifacts, black squares that flicker in desktop and 2D mode

[ThirtyNinety]

@ThirtyNinety
no trouble getting what this T is about, just stating that its funny to see ppl not trying all things possible, just to see if it makes a difference (or not),
just to make it easier to exclude or maybe pinpoint cause (easier).
never said anything about it being a (permanent) fix, nor that its to be accepting, when talking about a product purchased for money.

e.g., if i put 20L into a 100L car tank, i dont need to know its previous state, and still can make sure the engine has enough to start up..
im all about trouble shooting/finding the cause (for problems), just would like to see ppl do "ALL" offered tips,
even if its just to confirm (that it wasnt making any difference)..

that said, i havent messed with anything above 2xxx series, as i decided to wait for next gen.

I understand your suggestion but unfortunately, underclock has already been tried by many ppl here (as stated on the previous pages) and did not work.
As the problem only occurs in P-state 0 (= as in "rendering 3D at the appropriate clocks"), for me, the troubleshooting now shifted towards locking this state immediately at startup. As this is the only moment when I see these glitches occasionally...

If this does not work (= neither to avoid the rare problem, nor to reproduce it by locking P8-state at startup f.i.), I still have some other ideas, just wanna try them one by one. But as it can take months until I see the artifact again, I'm not sure if I will find something working forever, before already moving on to a next build...

 

[Waldorf]

did you try a (slight) vram underclocking, say ~25mhz?

 

[ThirtyNinety]

did you try a (slight) vram underclocking, say ~25mhz?

I will, but only after the current scenario where I'm now still testing whether forcing full gpu load during windows boot makes any difference for the artifacting. If it doesn't, I will test the opposite.

 

[Waldorf]

 

[CDSK]

Whelp, I'm now a proud member of this thread, not very stoked. Just wanted to share for the sake of cutting down on the rarity, since I didn't seem to find much information out there.

I know nothing about the technical side of things, so most of this thread is way above my head, just doing my due diligence.

Had my GPU (3070ti FE) for about a year and change, never mined or put it under a heavy load at all. Never OCed, temps run fine, A1 revision. It happened from a cold boot first thing in the morning and, as another user stated, the artifacting (exactly like this one) showed in a browser (Firefox, playing Youtube) while alt+tabbing between it and a game. While the game was absolutely fine, once closed a couple artifacts showed on desktop (so both monitors, ruling out bad cable/monitor). I did happen to notice that Windows pulled a stealth update on me recently, as well.

So far, I've only cleaned re-installed my driver and restarted... haven't seen them back, but I'm paranoid.

 

[mclaren85]

Today my GPU artifacts have reached yet another level. Right after the restart (after installing totally unrelated driver -" realtek sound driver") there were colorful stripes all around the screen. The stripes were merely a blend of what the current screen was.. The stripes were gone after the next restart; however, in order to relieve myself, I've run a full Furmark test which turned out to be all good. So my question is could a successful furmark test result rule out any gpu defect?

 

[ThirtyNinety]

First thing always to do here is ruling out whether the nvidia graphical driver was also acting weird, or crashing at this point by checking logs et cetera. If no software can be indicated as culprit, chances are more likely it’s a hardware thing, unfortunately. Furmark is not really reliable as a gpu hardware check, I’d always advise intensive tests with a Mods/Mats usb stick.

 

[Maxsenza151]

I think this is one of the worst cases.. my screen is literally going crazy sometimes, and some games are crashing also.
My gpu is RTX3080 drivers 531.41

 

[lordofrandy]

I think this is one of the worst cases.. my screen is literally going crazy sometimes, and some games are crashing also.
My gpu is RTX3080 drivers 531.41

I have the original black square artifacting talked about in this thread, and I have also had the artifacts that you show in the second image. My discord app was showing those triangle looking things and disappeared on reboot.

 

[ThirtyNinety]

You can trigger the same symptom with G-Sync enabled in the driver with a compatible monitor and V-sync enabled in a directX 12 (yes not 11) game when the vram clocks are fluctuating quite a bit. I managed to replicate and record this scenario in a new game to be released soon, but cannot share footage which proves this yet because of an embargo. Will share it afterwards. Also found the cause for this in my particular scenario and workaround for it which implicates the thing I already mentioned: overclock gpu to the supported max values for core and mem or, less preferred, enable performance mode in the nvidia driver settings.

 

[SashoDeBoy]

Try plugging into another outlet or another room or try a different house.
Thumb me up if this worked - I read the first page but didn't read the full thread as my dog needs a walk!

 

[lordofrandy]

You can trigger the same symptom with G-Sync enabled in the driver with a compatible monitor and V-sync enabled in a directX 12 (yes not 11) game when the vram clocks are fluctuating quite a bit. I managed to replicate and record this scenario in a new game to be released soon, but cannot share footage which proves this yet because of an embargo. Will share it afterwards. Also found the cause for this in my particular scenario and workaround for it which implicates the thing I already mentioned: overclock gpu to the supported max values for core and mem or, less preferred, enable performance mode in the nvidia driver settings.

you recreated the black squares on 2d mode?

 

[ThirtyNinety]

you recreated the black squares on 2d mode?

Yes. More specific, when switching between 3D and 2D with alt tabbing within a particular game. Is this representative for my own artifacting problem on startup? Nope not really, it isn’t, as that is always also accompanied by a Nvidia driver crash. But at least it learned me something new again. To be continued, when the embargo is lifted.

 

[ThirtyNinety]

In this video, you'll see that a DirectX 12 game under certain conditions can cause the Nvidia driver to show artifacts too when switching between 2D and 3D.
This happened when enabling G-Sync (for fullscreen/windowed mode and running the game in this modes does not matter) on my IPS panel and when also enabling V-Sync in-game.
Based on the phenomena, people would often say here, "BAD GPU or VRAM" but this is not the case here at all.

Most interesting for me however, is the amount of possible workarounds:

• G-Sync could be disabled in the Nvidia Control panel (so change on the driver operational level)
• V-Sync could be disabled in the game itself (which I preferred)
• I could also start recording the desktop and game to remove these artifacts
• I could also overclock the GPU slightly into P0-state territory, as shown in the video.

I have tested this game with another LCD panel which is also G-Sync-compatible and two other non-G-Sync panels combined with the same gpu and 3 other Nvidia RTX 30xx gpus, the same and different driver versions and in no other scenario, these artifacts appeared. So looks like a very specific scenario based on lots of variables to be able to reproduce this.

 

[ercarp]

Wow, I didn't think I'd actually be able to find a thread like this of so many people describing the exact issue I encountered today.

My computer and all of its components — including the RTX 3070 — are only about two months old and aside from this one strange issue I haven't had any problems whatsoever. 3DMark and all other benchmarks pass smoothly. Temperatures are great and I haven't overclocked anything.

Earlier today after starting up my computer, I had these small black boxes form near the middle of my screen. On both monitors, in the exact same spot. Shortly after, my screen flashed black and I got some nvlddmkm error in my Event Viewer. The black boxes were still there so I restarted my computer and lo and behold, the artifacts were gone. I tried to restart my PC multiple times afterwards to replicate the artifacting, but they never came back.

 

[ThirtyNinety]

Wow, I didn't think I'd actually be able to find a thread like this of so many people describing the exact issue I encountered today.

My computer and all of its components — including the RTX 3070 — are only about two months old and aside from this one strange issue I haven't had any problems whatsoever. 3DMark and all other benchmarks pass smoothly. Temperatures are great and I haven't overclocked anything.

Earlier today after starting up my computer, I had these small black boxes form near the middle of my screen. On both monitors, in the exact same spot. Shortly after, my screen flashed black and I got some nvlddmkm error in my Event Viewer. The black boxes were still there so I restarted my computer and lo and behold, the artifacts were gone. I tried to restart my PC multiple times afterwards to replicate the artifacting, but they never came back.

That is exactly the (=my) issue. But working on that. It's certainly a software issue, people who still dare to argue that now, only lack the knowledge about the specific circumstances. Just as we, the victims, do. Just enjoy your games and once it happens again, just reboot the pc, for now.

 

[ercarp]

That is exactly the (=my) issue. But working on that. It's certainly a software issue, people who still dare to argue that now, only lack the knowledge about the specific circumstances. Just as we, the victims, do. Just enjoy your games and once it happens again, just reboot the pc, for now.

I don't know if it helps, but my monitor setup is as follows: main monitor 165Hz IPS plugged in with DP, GSYNC enabled; second monitor 75Hz plugged in with HDMI, does not support GSYNC.

I've also read about a lot of games having issues with DX12, especially in conjunction with the hardware accelerated GPU scheduling feature Win11 has. Not sure if any of that relates to this though, but I thought I would mention it anyway.

Stressful times.

 

[ThirtyNinety]

I don't know if it helps, but my monitor setup is as follows: main monitor 165Hz IPS plugged in with DP, GSYNC enabled; second monitor 75Hz plugged in with HDMI, does not support GSYNC.

I've also read about a lot of games having issues with DX12, especially in conjunction with the hardware accelerated GPU scheduling feature Win11 has. Not sure if any of that relates to this though, but I thought I would mention it anyway.

Stressful times.

I can already tell you this is unrelated. I also have a G-Sync display and a second FreeSync-only display attached (also Display port 1.4a both), but did already see the artifacts without having that monitor attached. In short, the driver will not crash because of some attached monitor, at least not in my case. The same goes for HAGS, the Windows settings around VRR, Nvidia driver version and settings, and the OS as well: I have the artifacts on windows 10 and 11 dual-boot with both up to date drivers.

I must say, this is the weirdest issue I have ever seen in my 22 years of IT experience. So randomly happening. What are the variables, and where is the constant. Really want to get to the bottom with this.

 

[Pekmez Pita]

Yep.

I just recently started having the exact same issue with my brand new ASUS 3070 TI Tuf that came with my prebuilt that i bought in February.

Just like some others already have mentioned, i will randomly run into desktop artifacting and driver crashes (nvlddmkm) almost immediately after a cold boot without even stressing the system or running any programs at all. And yes, rebooting the system resolves the issue for me too. Started about two weeks ago for me. It happens infrequently but it happens. I initially thought it was happening because i had the GPU slightly undervolted but that didn't turn out to be case.

I have stressed the system with benchmarks and games but i can't replicate the issue. Changed drivers, disabled hardware acceleration in everything (browser, steam etc.), turned off hardware gpu scheduling and so on. Like i said. Only happens on cold boot and it appears to be random. That's all i know. Really strange issue indeed.

Appreciate the effort and insight, ThirtyNinety. I wish Nvidia had your work ethic.

Anyway. I am still eligible for a refund where i bought the prebuilt. So i think i am gonna return the system entirely since RMA'ing the card is unlikely to fix this issue as evidenced by others. Nvidia is clearly not concerned about this particular issue.


And, no. The card is not overheating either. It idles at 26-28C and 66-68C under full load.

 

[ThirtyNinety]

Great news today.

It is NOT the driver (or at least, not an sich)
and NO it is not the hardware.

Do not have a cause yet, but still working on a system which has the artifact issues now, managed to workaround the usual freezing / BSOD'ing and managed to fix the artifacts without rebooting. More information will follow.

Edit 1:
Avoiding to create a long story this time, I can confirm that DWM.EXE, Desktop Window Manager of the OS, is 100% sure the culprit.
Any setting in the nvidia driver or nvidia executable that I could kill, even killing explorer.exe, did not fix the artifacts. However, killing dwm.exe did that immediately.
After killing it, of course I needed to signout using Ctrl + Alt + Del --> Signout. Also after re-login, all artifacts stayed gone and now already for hours working in that session without seeing any artifact again. However, these can still reoccur right after the usual driver timeout, as long as I don't do anything to fix it.

That brings me to this topic: MultiPlane Overlay

Microsoft, AMD, Nvidia, are all sleeping on Windows 11 MPO display issues

Online forums are filled with users complaining about various display issues like stuttering, flickering, and white screens on Windows. A common suspect in many of these seems to be MPO.

www.neowin.net

[Feature Request] Disable Multiplane Overlay (MPO)

Inspired by horrendous 2 months of troubleshooting why my browser and everything changes colors when subject to anything such as a video on a Twitter or Facebook timeline or anything that uses the same codec(?), DDU reinstalls, cable switching, reformats, I would love to have that as a tickbox...

www.techpowerup.com

I will also disable MPO now, to see whether detaching MPO (as a subcomponent) from DWM fixes it permanently, I always game in Fullscreen if possible anyway.

Edit 2: can already confirm that the artifacts in Minecraft Legends (see previous posts) are immediately gone after disabling MPO and playing it Windowed + V-Sync in-game enabled, with all the rest untouched: same driver version, same game version, same Windows settings, except the MPO one.
I don't think disabling MPO is a real solution, as it's always better to have MPO in a working state, but at least we need something to point out as a cause because DWM is interwoven with so many things.

Edit 3:
Keep in mind that the dxdiag.txt log file where the MPO status can be checked is very misleading, as it is based on the (primary?) display. When I only leave my FreeSync monitor connected, which is not G-Sync compatible, MPO is effectively disabled but with the other monitor attached, it seems enabled. Just have to test now whether this only affects the logging, or really the functionality too on my G-Sync compatible display.

 

[lordofrandy]

Great news today.

It is NOT the driver (or at least, not an sich)
and NO it is not the hardware.

Do not have a cause yet, but still working on a system which has the artifact issues now, managed to workaround the usual freezing / BSOD'ing and managed to fix the artifacts without rebooting. More information will follow.

Edit 1:
Avoiding to create a long story this time, I can confirm that DWM.EXE, Desktop Window Manager of the OS, is 100% sure the culprit.
Any setting in the nvidia driver or nvidia executable that I could kill, even killing explorer.exe, did not fix the artifacts. However, killing dwm.exe did that immediately.
After killing it, of course I needed to signout using Ctrl + Alt + Del --> Signout. Also after re-login, all artifacts stayed gone and now already for hours working in that session without seeing any artifact again. However, these can still reoccur right after the usual driver timeout, as long as I don't do anything to fix it.

That brings me to this topic: MultiPlane Overlay

Microsoft, AMD, Nvidia, are all sleeping on Windows 11 MPO display issues

Online forums are filled with users complaining about various display issues like stuttering, flickering, and white screens on Windows. A common suspect in many of these seems to be MPO.

www.neowin.net

[Feature Request] Disable Multiplane Overlay (MPO)

Inspired by horrendous 2 months of troubleshooting why my browser and everything changes colors when subject to anything such as a video on a Twitter or Facebook timeline or anything that uses the same codec(?), DDU reinstalls, cable switching, reformats, I would love to have that as a tickbox...

www.techpowerup.com

I will also disable MPO now, to see whether detaching MPO (as a subcomponent) from DWM fixes it permanently, I always game in Fullscreen if possible anyway.

Edit 2: can already confirm that the artifacts in Minecraft Legends (see previous posts) are immediately gone after disabling MPO and playing it Windowed + V-Sync in-game enabled, with all the rest untouched: same driver version, same game version, same Windows settings, except the MPO one.
I don't think disabling MPO is a real solution, as it's always better to have MPO in a working state, but at least we need something to point out as a cause because DWM is interwoven with so many things.

Edit 3:
Keep in mind that the dxdiag.txt log file where the MPO status can be checked is very misleading, as it is based on the (primary?) display. When I only leave my FreeSync monitor connected, which is not G-Sync compatible, MPO is effectively disabled but with the other monitor attached, it seems enabled. Just have to test now whether this only affects the logging, or really the functionality too on my G-Sync compatible display.

Next time I see these artifacts on boot I will stop dwm.exe and report back.

 

[Supremacy]

3080 FE. Happened on 3 occasions.
The first it stopped after a few moments. 2nd, one month later it happened slightly more aggresively that led to a BSOD with "Video Scheduler Internal Error". After that i complete uninstalled driver with DDU. Everything had been fine for around one week until I was ironically watching a gpu video on HW Unboxed and get a batch of artifacts within the player. (not fullscreen)

Does not happen in games only Desktop and Browser.
(But to be fair I haven't been playing much games lately)

My GPU has no odd performance or Temps. It is not OC'd, infact it is only undervolted. I have my Windows Power Setting on Balanced and Nvidia Power Managment on Normal, so my GPU use is in a very low level state whenever it is idle.

Happening ever since RTX Video Super Sampling Driver if i'm remembering correctly.

Respect @ThirtyNinety good luck with your findings, so long as it's not hardware I'm relieved. Don't have money to get a gpu and i assume RMA is long over for 30s. The troubling part for me is that it actually led to a BSOD which has not happened to necessarily everyone.

 

[Sora]

3080 FE. Happened on 3 occasions.
The first it stopped after a few moments. 2nd, one month later it happened slightly more aggresively that led to a BSOD with "Video Scheduler Internal Error". After that i complete uninstalled driver with DDU. Everything had been fine for around one week until I was ironically watching a gpu video on HW Unboxed and get a batch of artifacts within the player. (not fullscreen)

Does not happen in games only Desktop and Browser.
(But to be fair I haven't been playing much games lately)

My GPU has no odd performance or Temps. It is not OC'd, infact it is only undervolted. I have my Windows Power Setting on Balanced and Nvidia Power Managment on Normal, so my GPU use is in a very low level state whenever it is idle.

Happening ever since RTX Video Super Sampling Driver if i'm remembering correctly.

Respect @ThirtyNinety good luck with your findings, so long as it's not hardware I'm relieved. Don't have money to get a gpu and i assume RMA is long over for 30s. The troubling part for me is that it actually led to a BSOD which has not happened to necessarily everyone.

This issue has no BSOD occurences or causing games to crash, its purely an issue on the output side of things at the display engine / scaler when hardware overlays are being applied, and only when a DSC stream is active.

So posts like....

I think this is one of the worst cases.. my screen is literally going crazy sometimes, and some games are crashing also.
My gpu is RTX3080 drivers 531.41

Are demonstrating a completely different issue.

I thought i had already posted to this thread, but it might have been another site, this issue is only so far known to happen on displays utilising DSC to achieve high refresh rates or resolutions, at 10bpc, and is only a visual flaw.

In this video, you'll see that a DirectX 12 game under certain conditions can cause the Nvidia driver to show artifacts too when switching between 2D and 3D.
This happened when enabling G-Sync (for fullscreen/windowed mode and running the game in this modes does not matter) on my IPS panel and when also enabling V-Sync in-game.
Based on the phenomena, people would often say here, "BAD GPU or VRAM" but this is not the case here at all.

Most interesting for me however, is the amount of possible workarounds:

• G-Sync could be disabled in the Nvidia Control panel (so change on the driver operational level)
• V-Sync could be disabled in the game itself (which I preferred)
• I could also start recording the desktop and game to remove these artifacts
• I could also overclock the GPU slightly into P0-state territory, as shown in the video.

I have tested this game with another LCD panel which is also G-Sync-compatible and two other non-G-Sync panels combined with the same gpu and 3 other Nvidia RTX 30xx gpus, the same and different driver versions and in no other scenario, these artifacts appeared. So looks like a very specific scenario based on lots of variables to be able to reproduce this.

In that video, the display engine is fighting over the plane order with MPO enabled, and given your other posts in this thread, you have a different issue too, the black blocks are entirely benign and do not affect stability at all, nor do they log an error into event viewer.

You may have the other issue currently known in the drivers, where High VRAM loads across multiple applications introduces instability and events are logged as applications are closed down.

Hmm not needed to reboot imo as you can just do this with Win+Ctrl+Shift+B

That command issues a reset command to the driver while it is still loaded, but it does not work, in fact, I don't think that command has ever worked correctly.

This key combo does not nor is it designed to cause a driver reset, it can result in one if the tdr timeout is hit in the process of reinvoking the display engine, but its behavior as implemented is to perform a mode swap and refresh the image in the case a display has resumed with a black screen, and then immediately write out and submit a black screen diagnostics event log to microsoft.


AND OPS ISSUE HAS NOTHING TO DO WITH VRAM STABILITY.
so the people chiming in with issues that are appearing on top of 3d games, and did not coincide to an upgrade from 472, can stop participating in the thread because they do not have OPS issue, and are mixing another, possibly unique to a faulty card issue in with the ops.

@BenchAndGames did you ever make those kernel dumps and submit them to nvidia as instructed elsewhere?

 

[ThirtyNinety]

If OP his issue has nothing to do with stability (related to what you said in the middle of your text, I am not referring to the vram stability), why then can I find this in the very first post :

The issue is very random, there is no chance to reproduce it, the funny fact is that it will have everytime same patron I mean everytime there are some squares that are flickering in desktop, if you move mouse they dissapear, apear again, flickering and most likely the nvidia driver will crash with display has stopped working and recovered. They thend to flicker at same spot every time. Here are some examples of the issue that I found from other users but all of them with RTX 3000 series cards.

And that one is logged in the event viewer too, right?
OP never said the issue is only a visual flaw.
And nope my issue is not related to high VRAM loads, it is only happening in 2D desktop as stated here many times.

Anyway, with that, I think we are now back for some time with vague theories about faulty cards or drivers I guess.

 

[Sora]

And that one is logged in the event viewer too, right?

Then the OP has mixed 2 issues up in here as well, the black squares issue was first reported by Venturi on guru3d and no driver crash aspect has ever been linked to it by him or other users reporting the issue on drivers 496.13 and newer.

Actually, Venturi found a workaround in disabling the nvidia driver helper service unless absolutely needed for the CPL