Youtube, Discord, Visual Studio Code are flickering with black checkered squares/vertical lines. Partial solutions to this issue.

[chaoshusky]

Well, for the first time ever today i've had a similar issue when on the login screen, then after logging in.. However the strangeness disappears if i close or minimize and re-open and it hasn't happened again. Was from a cold boot today...hmm.. I hope my 3090 isn't dying!

No I'm pretty sure the OP is right that this only started with the 2022H2 update for Win11. So it's likely a buggy interaction between the nvidia driver and the OS after that update.

I've only just had this happen, so its likely just a driver regression somewhere. I was worried for a brief moment my VRAM was dying but..seems pretty random so far! Only happened once, today, which is strange.

 

[KsushaTeaKisa]

Well, for the first time ever today i've had a similar issue when on the login screen, then after logging in.. However the strangeness disappears if i close or minimize and re-open and it hasn't happened again. Was from a cold boot today...hmm.. I hope my 3090 isn't dying!


I've only just had this happen, so its likely just a driver regression somewhere. I was worried for a brief moment my VRAM was dying but..seems pretty random so far! Only happened once, today, which is strange.

Your GPU is absolutely fine, don't worry

 

[Selaya]

No I'm pretty sure the OP is right that this only started with the 2022H2 update for Win11. So it's likely a buggy interaction between the nvidia driver and the OS after that update.

i've experienced this behavior since i first started using 7 in like, 2009
i've seen those kind of artifacts; admittedly not as often/gamebreaking but this issue's been around since then.

 

[KsushaTeaKisa]

i've experienced this behavior since i first started using 7 in like, 2009
i've seen those kind of artifacts; admittedly not as often/gamebreaking but this issue's been around since then.

Apparently these "artifacts" are related to Nvidia rendering scenes starting with nv maxwell (tile based), this is approximately 2014-2015, but not 2009.
Also, all the people who have encountered this bug claim that it began in the fall of last year, exactly after 22H2 release for Windows 10/11.

Therefore, most likely you saw something similar to this bug, but it was definitely different.

 

[Aquinus]

This sounds like a GPU driver issue or an issue with the GPU offload in Chome/Electron apps. Try disabling GPU acceleration in chrome and try again. If it works fine, use a driver cleaner and reinstall your GPU drivers.

 

[Zer0code]

This sounds like a GPU driver issue or an issue with the GPU offload in Chome/Electron apps. Try disabling GPU acceleration in chrome and try again. If it works fine, use a driver cleaner and reinstall your GPU drivers.

Reinstalling the driver and even reinstalling Windows 10 / 11 will not solve the issue!

Disabling GPU acceleration fixes the problem. But this is a bad solution.
Why would I need an expensive GPU that can't do GPU acceleration properly in programs?

 

[Aquinus]

Disabling GPU acceleration fixes the problem. But this is a bad solution.
Why would I need an expensive GPU that can't do GPU acceleration properly in programs?

This alone tells me that your GPU is either unstable or the driver isn't behaving the way it's supposed to. I suspect the latter. Try different versions of nVidia's drivers, this could be an issue with the 1000-series for this particular driver. I've never encountered a case where disabling gpu acceleration solving it didn't point to an issue with the driver. To be honest, it wouldn't be the first time that a particular driver version caused some kind of regression. Try either an older driver or if nVidia has a beta, you can always try that. It's also worth checking the release notes for known issues.

 

[alexboliam]

This alone tells me that your GPU is either unstable or the driver isn't behaving the way it's supposed to. I suspect the latter. Try different versions of nVidia's drivers, this could be an issue with the 1000-series for this particular driver. I've never encountered a case where disabling gpu acceleration solving it didn't point to an issue with the driver. To be honest, it wouldn't be the first time that a particular driver version caused some kind of regression. Try either an older driver or if nVidia has a beta, you can always try that. It's also worth checking the release notes for known issues.

Seems like you didn't read the whole thread or the OP's post. NVIDIA itself confirmed that this is a bug and added it to the official bug list.

 

[KsushaTeaKisa]

This alone tells me that your GPU is either unstable or the driver isn't behaving the way it's supposed to. I suspect the latter. Try different versions of nVidia's drivers, this could be an issue with the 1000-series for this particular driver. I've never encountered a case where disabling gpu acceleration solving it didn't point to an issue with the driver. To be honest, it wouldn't be the first time that a particular driver version caused some kind of regression. Try either an older driver or if nVidia has a beta, you can always try that. It's also worth checking the release notes for known issues.

Read the previous comments in this thread, I already said that Nvidia has confirmed this bug. And yes, this issue is not solved by reinstalling the drivers. This issue is related between the 22H2 update and any drivers of Nvidia.

 

[ThirtyNinety]

No I'm pretty sure the OP is right that this only started with the 2022H2 update for Win11. So it's likely a buggy interaction between the nvidia driver and the OS after that update.

it's not, I had it already in Windows 10.
But the rest is true: it's something with the WDM.exe

Check this.

Also, setting Angle in chrome flags on OpenGL gives me a myriad of hickups in live video streams f.i., so had to set this back to Dx11.
Addition: these hickups may have been something with Chrome in general, because these also dared to occur later on in Dx11 angle, and were totally gone when switching to Firefox (without altering the defaults there)

 

[KsushaTeaKisa]

it's not, I had it already in Windows 10.
But the rest is true: it's something with the WDM.exe

Check this.

Also, setting Angle in chrome flags on OpenGL gives me a myriad of hickups in live video streams f.i., so had to set this back to Dx11.

Thanks for your experience.
Unfortunately, disabling MPO will not help with this bug. I used a disabled MPO for several months for tests and it only provoked more flickering (as in my video in this post. Frequent flickers are caused by the disabled MPO).
Now I prefer to use enabled MPO + HAGS. This reduces the occurrence of flickering to a minimum, and also flickering has appear less common in Discord and Visual Code. I think this is due to the fact that MPO and HAGS add some load to the GPU at certain points, and as it turned out, these artifacts may appear when the GPU is in low P-State (P8) and related to DirectComposition Video Overlays (as confirmed m_w_h from Nvidia Reddit).

By the way, I'm not surprised that the DWM is related to the problem from your topic. After the 22H2 update, there were problems with the DWM, which sudden GPU spikes up to 100% if HAGS was enabled. As far as I know this problem still exists even if HAGS is turned off for some GPUs. This all suggests that after 22H2 there has serious problems with the compatibility of the OS and Nvidia drivers, and it's bad that Microsoft still doesn't know about it.
Also, I must say that after 22H2, AMD GPUs also started having some problems when MPO is enabled You may have already noticed this when you searched for posts with MPO.

P.S: epic bug in Minecraft. I 'm glad you found a way to fix this. I also play Minecraft Legends / Minecraft for Windows 10 / Minecraft Java, but I don't have this "rendering artifacts".

 

[ThirtyNinety]

Thanks for your experience.
Unfortunately, disabling MPO will not help with this bug. I used a disabled MPO for several months for tests and it only provoked more flickering (as in my video in this post. Frequent flickers are caused by the disabled MPO).
Now I prefer to use enabled MPO + HAGS. This reduces the occurrence of flickering to a minimum, and also flickering has appear less common in Discord and Visual Code. I think this is due to the fact that MPO and HAGS add some load to the GPU at certain points, and as it turned out, these artifacts may appear when the GPU is in low P-State (P8) and related to DirectComposition Video Overlays (as confirmed m_w_h from Nvidia Reddit).

By the way, I'm not surprised that the DWM is related to the problem from your topic. After the 22H2 update, there were problems with the DWM, which sudden GPU spikes up to 100% if HAGS was enabled. As far as I know this problem still exists even if HAGS is turned off for some GPUs. This all suggests that after 22H2 there has serious problems with the compatibility of the OS and Nvidia drivers, and it's bad that Microsoft still doesn't know about it.
Also, I must say that after 22H2, AMD GPUs also started having some problems when MPO is enabled You may have already noticed this when you searched for posts with MPO.

Thanks for the fast feedback, that will save me a lot of needless time then I guess
In each case, I can also tell you that P-state does not matter that much. Today I had the "luck" to be able to test stuff while having the artifacts for hours, created a log for each step and killing dwm.exe was the only thing to avoid the reboot which would remove the ability for further troubleshooting.

Very interesting stuff you are referring to, but unfortunately:

• I don't believe the issue is limited to Windows 11 22H2 because I have seen this already in Windows 10 and earlier Windows 11 builds
• and also not limited to chromium apps only, as i always log at startup which apps are using the gpu with nvidia-smi -q -d PIDS and still, when no chromium apps are in the array of opened applications, the artifacts can appear.
  
  Normally I only have these 4 in the list, yes including DWM.

    Processes
        Process ID                        : 1548
            Type                          : C+G
            Name                          : C:\Windows\System32\dwm.exe
            Used GPU Memory               : Not available in WDDM driver model
        Process ID                        : 5208
            Type                          : C+G
            Name                          : C:\Windows\explorer.exe
            Used GPU Memory               : Not available in WDDM driver model
        Process ID                        : 8276
            Type                          : C+G
            Name                          : C:\Windows\SystemApps\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\SearchHost.exe
            Used GPU Memory               : Not available in WDDM driver model
        Process ID                        : 8340
            Type                          : C+G
            Name                          : C:\Windows\SystemApps\Microsoft.Windows.StartMenuExperienceHost_cw5n1h2txyewy\StartMenuExperienceHost.exe
            Used GPU Memory               : Not available in WDDM driver model

 

[KsushaTeaKisa]

Thanks for the fast feedback, that will save me a lot of needless time then I guess
In each case, I can also tell you that P-state does not matter that much. Today I had the "luck" to be able to test stuff while having the artifacts for hours, created a log for each step and killing dwm.exe was the only thing to avoid the reboot which would remove the ability for further troubleshooting.

Very interesting stuff you are referring to, but unfortunately:

• I don't believe the issue is limited to Windows 11 22H2 because I have seen this already in Windows 10 and earlier Windows 11 builds
• and also not limited to chromium apps only, as i always log at startup which apps are using the gpu with nvidia-smi -q -d PIDS and still, when no chromium apps are in the array of opened applications, the artifacts can appear.
  
  Normally I only have these 4 in the list, yes including DWM.

    Processes
        Process ID                        : 1548
            Type                          : C+G
            Name                          : C:\Windows\System32\dwm.exe
            Used GPU Memory               : Not available in WDDM driver model
        Process ID                        : 5208
            Type                          : C+G
            Name                          : C:\Windows\explorer.exe
            Used GPU Memory               : Not available in WDDM driver model
        Process ID                        : 8276
            Type                          : C+G
            Name                          : C:\Windows\SystemApps\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\SearchHost.exe
            Used GPU Memory               : Not available in WDDM driver model
        Process ID                        : 8340
            Type                          : C+G
            Name                          : C:\Windows\SystemApps\Microsoft.Windows.StartMenuExperienceHost_cw5n1h2txyewy\StartMenuExperienceHost.exe
            Used GPU Memory               : Not available in WDDM driver model

Since I know how to reproduce these artifacts on my PC (using some videos from Youtube 2K / 4K + dark ambient theme) this month I tried previous versions of Windows 10 (20H2, 21H1, 21H2) and there were no artifacts anywhere, but only by installing 22H2 I again started to see it, so I just once again made sure that this bug has a related with 22H2 update.

However, there is good news: Nvidia has officially confirmed this devil bug from this thread, so now we need to wait for them to fix it as soon as possible. Most likely it will be the same as with Forza 4, when game were crash each 10-15 min due to the incompatibility of the latest Nvidia drivers and Windows, and then Nvidia finally contacted Microsoft after 5-6 months and they released a joint update in the game

P.S: Initially, I started seeing these artifacts when using Windows 10, even on my video you can see Windows 10, in this topic I did not claim that this issue only affects Windows 11

 

[ThirtyNinety]

Since I know how to reproduce these artifacts on my PC (using some videos from Youtube 2K / 4K + dark ambient theme) this month I tried previous versions of Windows 10 (20H2, 21H1, 21H2) and there were no artifacts anywhere, but only by installing 22H2 I again started to see it, so I just once again made sure that this bug has a related with 22H2 update.

However, there is good news: Nvidia has officially confirmed this devil bug from this thread, so now we need to wait for them to fix it as soon as possible. Most likely it will be the same as with Forza 4, when game were crash each 10-15 min due to the incompatibility of the latest Nvidia drivers and Windows, and then Nvidia finally contacted Microsoft after 5-6 months and they released a joint update in the game

P.S: Initially, I started seeing these artifacts when using Windows 10, even on my video you can see Windows 10, in this topic I did not claim that this issue only affects Windows 11

Okay then, let's hope for the best soon, thanks for all the effort!

For now I will stay on MPO disabled, just to learn for myself what is different.
Actually still looking a way to see it is actually disabled as exported dxdiag.txt still says I have the 4 planes and don't know yet whether Special K actually exposes that information

 

[KsushaTeaKisa]

Okay then, let's hope for the best soon, thanks for all the effort!

For now I will stay on MPO disabled, just to learn for myself what is different.
Actually still looking a way to see it is actually disabled as exported dxdiag.txt still says I have the 4 planes and don't know yet whether Special K actually exposes that information

When I disabled MPO my result of DXDiag was:
- MPO MaxPlanes: 1
- MPO Caps: Not Supported
- MPO Stretch: Not Supported
- MPO Media Hints: Not Supported
- MPO Formats: Not Supported

However, when I tried to restore this MPO back, the result of DxDiag did not change and remained the same. The only thing that worked for me was a clean install of Windows. Now the same problem exists on my second PC but on Windows 11, I can't disable MPO and the result is the same as yours. This is very strange, so be aware that you may have the same problem.
I just want to make sure you are using the files from here?
I hope you'll manage to disable MPO without time-consuming reinstallation of Windows. Share your experience of using a disabled MPO in the future

 

[ThirtyNinety]

When I disabled MPO my result of DXDiag was:
- MPO MaxPlanes: 1
- MPO Caps: Not Supported
- MPO Stretch: Not Supported
- MPO Media Hints: Not Supported
- MPO Formats: Not Supported

However, when I tried to restore this MPO back, the result of DxDiag did not change and remained the same. The only thing that worked for me was a clean install of Windows. Now the same problem exists on my second PC but on Windows 11, I can't disable MPO and the result is the same as yours. This is very strange, so be aware that you may have the same problem.
I just want to make sure you are using the files from here?
I hope you'll manage to disable MPO without time-consuming reinstallation of Windows. Share your experience of using a disabled MPO in the future

Yes I used that reg-file, ran it as admin.
Indeed, that's what I was already afraid for. It seems like MPO already cannot disabled anymore as I nowhere see these "Not Supported" lines in the txt.

MPO MaxPlanes: 4
MPO Caps: RGB,YUV,BILINEAR,HIGH_FILTER,STRETCH_YUV,STRETCH_RGB,IMMEDIATE,HDR (MPO3)
MPO Stretch: 10.000X - 0.500X
MPO Media Hints: resizing, colorspace Conversion
MPO Formats: NV12,P010,YUY2,R16G16B16A16_FLOAT,R10G10B10A2_UNORM,R8G8B8A8_UNORM,B8G8R8A8_UNORM

But maybe lucky me, because this means I will not have to reinstall Windows when the issue is fixed (ever). No actually that is a lie, I take Veeam Endpoint Backups at a weekly base
For now, I will just kill dwm.exe and do a signout, or much easier, just reboot the pc.

Because I think I am almost nearing the end of all possibilities on end user side.
I've spent more than 6 hours on the issue today, just to only conclude it is certainly dwm.exe-related, and that the best programs to replicate it are chrome.exe and outlook.exe.

It's indeed what I thought, MPO cannot be disabled anymore.
You can test this by creating a C++ app which uses D3DKMTGetMultiPlaneOverlayCaps() f.i., it will report MPO as being supported despite having the OverlayTestMode registry key set.
Which brings me not a step further.
Luckily, you still can - See below

 

[KsushaTeaKisa]

Yes I used that reg-file, ran it as admin.
Indeed, that's what I was already afraid for. It seems like MPO already cannot disabled anymore as I nowhere see these "Not Supported" lines in the txt.

MPO MaxPlanes: 4
MPO Caps: RGB,YUV,BILINEAR,HIGH_FILTER,STRETCH_YUV,STRETCH_RGB,IMMEDIATE,HDR (MPO3)
MPO Stretch: 10.000X - 0.500X
MPO Media Hints: resizing, colorspace Conversion
MPO Formats: NV12,P010,YUY2,R16G16B16A16_FLOAT,R10G10B10A2_UNORM,R8G8B8A8_UNORM,B8G8R8A8_UNORM

But maybe lucky me, because this means I will not have to reinstall Windows when the issue is fixed (ever). No actually that is a lie, I take Veeam Endpoint Backups at a weekly base
For now, I will just kill dwm.exe and do a signout, or much easier, just reboot the pc.

Because I think I am almost nearing the end of all possibilities on end user side.
I've spent more than 6 hours on the issue today, just to only conclude it is certainly dwm.exe-related, and that the best programs to replicate it are chrome.exe and outlook.exe.

It's indeed what I thought, MPO cannot be disabled anymore.
You can test this by creating a C++ app which uses D3DKMTGetMultiPlaneOverlayCaps() f.i., it will report MPO as being supported despite having the OverlayTestMode registry key set.
Which brings me not a step further.

The strangest thing is that I just now tried to disable MPO on my first PC with a clean Windows 11 that I installed last week and my MPO became disabled. I'm afraid to see if I can restore it back...



Edit: okie dokie, I was able to restore MPO. I really didn't want to do a clean install of Windows again.

 

[ThirtyNinety]

The strangest thing is that I just now tried to disable MPO on my first PC with a clean Windows 11 that I installed last week and my MPO became disabled. I'm afraid to see if I can restore it back...

That's very weird indeed. I must admit that this is an upgraded Windows 10 to W11 build. Maybe some remnant blocks this setting for me?

 

[KsushaTeaKisa]

That's very weird indeed. I must admit that this is an upgraded Windows 10 to W11 build. Maybe some remnant blocks this setting for me?

Maybe, but I'm not sure about it. Because on my second PC there is also a clean installation of Windows and the same bug occurs there

 

[ThirtyNinety]

Maybe, but I'm not sure about it. Because on my second PC there is also a clean installation of Windows and the same bug occurs there

Aha, I found it. The dxdiag logging is just 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 now. Just don't know whether this only affects the logging, or really the functionality too.

Just need some kind of test scenario, where one can really see the difference between MPO enabled versus disabled ...

 

[KsushaTeaKisa]

Aha, I found it. The dxdiag logging is just 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 now. Just don't know whether this only affects the logging, or really the functionality too...

Good job!
The only thing I'm sure of is that DxDiag logs indicate several MPO values. Separately for the discrete graphics and for the integrated graphics, if you look at these logs from a laptop. Maybe this information will help you somehow

 

[ThirtyNinety]

Good job!
The only thing I'm sure of is that DxDiag logs indicate several MPO values. Separately for the discrete graphics and for the integrated graphics, if you look at these logs from a laptop. Maybe this information will help you somehow

It's a build with cpu without iGpu, with only 1 GPU, so should be fine.
But maybe you have a procedure to test MPO enabled versus disabled? Or a way to trigger the artifacts would also be nice, you said something about certain youtube vids and having a dark browser theme?

 

[KsushaTeaKisa]

It's a build with cpu without iGpu, with only 1 GPU, so should be fine.
But maybe you have a procedure to test MPO enabled versus disabled? Or a way to trigger the artifacts would also be nice, you said something about certain youtube vids and having a dark browser theme?

Yes, as I said in this post, each GPU model has its own videos where black checkered squares will appear or even flicker like in my video.
I can easily check if I have MPO and HAGS enabled or not, because if it is enabled, then those flickers on that video will not be as frequent as I showed in the post. Therefore, this method exists for my GPU model, as soon as Nvidia and Microsoft fix this bug with artifacts, I will be able to verify this very quickly.

I don’t know any other way to check if the MPO is working or not, sorry.

 

[KsushaTeaKisa]

I have new information about this bug that may be of interest to someone.

This bug has been around for over two years, I found a bug report about it in bugs.chromium.org. People from there have been trying to figure out how this bug works, they even maked "test videos" for it with an ambient background that resembles current Youtube's default ambient background in dark mode. But in the end, their investigation ended because the Chromium team did not respond to this bug report in any way. The autumn update of Windows 22H2 somehow expanded the threshold for this bug and it also began to appear in Discord, Visual Studio Code, Steam and apparently in all applications that use Electron. Also ambient background by default from Youtube only exacerbated the appearance of these artifacts when watching a video on Youtube. That's why most users like us encountered this issue only last fall.

What is known about this bug now:
1. This bug not only for Chromium, this is most likely an Electron issue that has a conflict with Nvidia drivers, because as I said artifacts appear everywhere Electron is using, based on my experience and reports from other users.
2. This only works on DirectX 11 API.
3. This is somehow related to the refresh rate display, because if I change my display mode from 75Hz to 60Hz, then in the videos where I have artifacts, they immediately disappear.
4. Windows 22H2 update really somehow expanded the threshold of this issue, because I can not reproduce this bug on other versions of Windows how I reproduced it before on my GPU model + as I said in the post, I began to see artifacts in Discord and Visual Studio Code immediately as soon as I updated to 22H2.
5. It also has something to related with DirectComposition Video Overlays as I mentioned in the comments earlier.
6. Artifacts may appear when the GPU is in low P-State (P8).

But fortunately recently this bug has been officially confirmed by Nvidia. Obviously, they are more likely to be able to find this issue in Electron, because they already know that there is a issue in Chromium and will definitely be able to figure out in this if Nvidia is not an indie studio.

Here is a link to that bug report, who wants to look into.

 

[Wezelkrozum]

Hi all,

Just a small correction to KsushaTeaKisa's notes:

1. This bug not only for Chromium, this is most likely an Electron issue that has a conflict with Nvidia drivers, because as I said artifacts appear everywhere Electron is using, based on my experience and reports from other users.

The Discord, VSCode and Steam apps use Electron. Electron uses Chromium. And the bug is in Chromium. That is why you see this bug in those apps. And you also see this bug in the Chrome, Opera and Edge browsers, because those use Chromium as well.

My current theory is that there might be something going wrong in Chromium's task scheduler that causes this bug when a video is playing while an additional graphical change is rendered on the webpage with a lower framerate than the display framerate. (Like the ambient light effect that is only rendered once per second or a few seconds.)
I think this goes wrong because Chromium's scheduler is finetuned for a display framerate of 60Hz. For example, when you have a monitor that runs at a higher refreshrate combined with those graphical changes, the video will play with stutters and framedrops. This happens because Chromium's task scheduler renders video frames to late sometimes: Chromium bug 1433124. So, I suspect that the incorrectly scheduled rendering of some video frame's in Chromium combined with the DirectComposition Video Overlays of the NVidia driver could sometimes set Chromium in a tricky state. A state where Chromium is decoding/rendering a new video frame while the NVidia driver is displaying the currently decoded/rendered video frame to the monitor. (This could also explain why you sometimes see black checkered artifacts and sometimes checkered artifacts that contain the previous video frame. Because some video's use i-frames and b-frames. I-frames are constructed from data in the previous video frame which could hint to those checkered artifacts that contain previous video frames. And b-frames are rendered from scratch which could hint to those black checkered artifacts. More info about video decoding can be found here: Video compression)

Anyway, because I've gathered some additional information via that other bug that I found last week, related to the checkered artifacts: Chromium bug 1433124 And because the reproduction of this bug seems to be the hardest hurdle at the moment. I'll (after 2,5 years ) once again attempt to create a solid test case this week. (But I won't promise that I will succeed).

PS: Some info about my setup:
GPU: NVidia RTX 2070 Super
CPU: Intel 6700K
Monitor: 1440p @ 144Hz