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

[alutyl]

I have tested with two different monitors (one from Viewsonic and one from Asus), Gsync both, without HDR... and the error always appears. Of course, I have also tried with different DP cables, from cheap to expensive.

I still think this is not a hardware problem...

Most likely because messages with screenshots of this problem began to appear about 3-4 months ago by several people at once.

 

[ThirtyNinety]

When the artifacts started appearing on the desktop here for the first time, at the same moment this was added in the event log:
System --> Event ID 13 from source "nvlddmkm"
\Device\Video3 Graphics Exception: ESR 0x404000=0x80000040

On other days, the Nvidia driver would just crash and TDR autorecover in a loop for a hundred times in a few seconds.
So I NEVER had the artifacts without also having an event log stating that something went wrong with the nvidia driver, and rebooting always fixed it.
I still must get to know a defective hardware part which is so quickly back in a fully operational status with a simple reboot.

Also important, the artifacts did not necessarily start to pop-up when opening something hardware-accelerated. Sometimes these were already present on the desktop upon boot, without opening any program.

In each case, the problem with artifacting is here gone now for 3.5 months, after changing a myriad of things.

 

[UltiZA]

So my last post was on March 9th... Since then I havent had a single artifact. I didn't change a single thing on my PC and have been using it for around 10h+ a day. I really don't understand whats going on

 

[Tarkovski]

So my last post was on March 9th... Since then I havent had a single artifact. I didn't change a single thing on my PC and have been using it for around 10h+ a day. I really don't understand whats going on

Congratulations! What Nvidia driver and W11 build are you currently using?

 

[alutyl]

My system configuration:
Ryzen 5 5600
Gigabyte B450 Aorus Elite V1
32GB RAM
Palit GeForce RTX 3070 GameRock V1

I have a problem similar to that of users Zer0code and Tarkovski. In my browser (google chrome) or in the discord sometimes appear artifacts. This never happens in games.
I also found a lot of people on reddit with the exact same artifacts. I don't know how to fix it. I have tested the memory of my graphics card in MATS/MODS. The test passed successfully. I ran cyberpunk on max settings to have 100% GPU utilization and 8gb memory utilization out of 8gb. I don't get any artifacts.
I also suspect that my monitor may somehow be causing these artifacts. I am using an LG 32QN600. It has HDR and FreeSync. Perhaps one of these technologies is causing the artifacts.
Zer0code and Tarkovski what can you say?
Here is a screenshot of my artifacts (don't look at the purple squares.).

Spoiler: What the artifacts look like on my computer.

View attachment 288526

I had both the first type of artifacts and the second type, which is on the screenshots from reddit.
(All these screenshots are from different people)

Spoiler: Here are some more screenshots of people from reddit

View attachment 288528

View attachment 288535

View attachment 288536
View attachment 288532

Spoiler: Here are some more screenshots of people from reddit.

View attachment 288533

View attachment 288529

Another person with the same artifacts as in my last post.

http://imgur.com/a/OI2gaVO

 

[Dr. Dro]

I have tested with two different monitors (one from Viewsonic and one from Asus), Gsync both, without HDR... and the error always appears. Of course, I have also tried with different DP cables, from cheap to expensive.

I still think this is not a hardware problem...

I do not use DisplayPort at all - HDMI connection here. The only time it has ever manifested was after a known-unstable overclock pushing benchmarks. It's not related to the cable, I can assure you of that 100%

Indeed, again this proves that this issue is not hardware-related.

Still strikes me as odd, I don't believe it's hardware either since my GPU is fine, it's gotta be something with the power delivery or memory timings with some setups. Adding to that it seems to occur exclusively in low-power pstates (idle to very low load), I'm inclined to believe that it's related to the memory timings.

I'm surprised no one found the exact root cause yet, I'm going by experience on my claims (and I have good reason to believe that it is the case - unstable VRAM somehow), but what kind of cards are peepos affected by the problem running? Might be helpful in pinpointing affected models and reporting to Nvidia somehow.

Mine's an ASUS TUF OC - bought launch day, so it's 100% the initial hardware revision if any came after that.

Congratulations! What Nvidia driver and W11 build are you currently using?

I haven't had this problem in over a year and I've run Windows 10 and 11's every build and almost every driver update since. Then again, that was the last time I was trying to chase 3DMark

 

[Zer0code]

I recorded another video. The color of the glitch depends on the background color of the site. This happens in both Opera and Edge browsers and only if there is no load on the video card.
Timecodes: 0:20, 0:46, 0:48, 1:06,

You can also see that some of the elements of the site flashing, it violates the rendering layer of the site. This is very strange!

I have a few suggestions:
1) It's a GPU defect.
2) It's an error at Nvidia driver level + some other driver.
3) It is a DirectX bug.
4) It may be motherboard and graphics card incompatibility.

 

[Hankieroseman]

I got one. After updating to nvidia's V.531.41 for new RTX4090 and Mandatory Update for MSFS2020. Doesn't do it anywhere else.Just MSFS2020. Not in DCS or youtube. Always same spot and never had one in RTX3090.

 

[Dr. Dro]

I recorded another video. The color of the glitch depends on the background color of the site. This happens in both Opera and Edge browsers and only if there is no load on the video card.
Timecodes: 0:20, 0:46, 0:48, 1:06,

You can also see that some of the elements of the site flashing, it violates the rendering layer of the site. This is very strange!

I have a few suggestions:
1) It's a GPU defect.
2) It's an error at Nvidia driver level + some other driver.
3) It is a DirectX bug.
4) It may be motherboard and graphics card incompatibility.

Safe to slash 3 and 4 off outright, IMO.

It's either a hidden design defect, or a driver bug/incorrect memory timings.

On another note: this has never happened on my laptop 3050 (GA107-type).

 

[ThirtyNinety]

Just had the artifacts too again, as usual after a cold startup this morning. Last time was on 6th of Dec 2022 with the previous driver version.
When I started recording them with Geforce Experience, they disappeared. Stopped the recording, they appeared seconds later again (on the recording also nothing special). Then some seconds later the screen became black and Windows 11 was started to get freezing, so I briefly pushed the power button and the pc succeeded just in time to turn off properly.

Again, in the event log as usual, first:
\Device\Video3
Graphics Exception: ILLEGAL_OPCODE

\Device\Video3
Graphics Exception: ESR 0x40a790=0x80000024

... and as a result of that, Windows TDR kicks in (you can see that this is only consequence, not cause, as it happens seconds later):
Display driver nvlddmkm stopped responding and has successfully recovered.

It started happening 5 minutes after booting up, and without opening any application/browser.
These actions were already performed one or many times, but did not fix the issue:
- Update GPU to latest driver after using DDU in safe mode
- Updated chipset to latest driver after using DDU in safe mode
- Updated mainboard to latest BIOS + loaded defaults
- Experimental Features of Geforce Experience disabled
- Hardware-accelerated GPU scheduling in Windows disabled/did not matter so afterwards re-enabled
- Disabled hardware acceleration individually when possible for all used applications
- iCue Enable Plugins disabled/re-enabled + deleted folders ASUS and NvidiaPlugin
- GPU Tweak 2/3 disabled/did not matter so afterwards re-enabled, with Gaming as default profile
- Set GPU in Line-Based Mode instead of default MSI using "E:\Bin\MSI_util_v3.exe" and rebooted/now reverted to MSI-mode
- Disabled/re-enabled Resizeable Bar in BIOS
- Moved GPU to another pci-e slot of the mainboard/Changed PCIe X16 slot Mode which should have options for Auto, Gen1, Gen2, Gen3 or Gen4. If its in Auto/Gen4, try to set it to Gen 3. If its in Gen 3, set try it to Auto
- Switch to other vBIOS when Dual vBios GPU, afterwards power off/reboot
- Disabled OC on CPU and RAM
- Memtest RAM: 0 errors
- Underclocked GPU
- Run GPU in Nvidia Control Panel Debug Mode (when supported) to reset it to reference clock speed: cannot be saved so no valid option for driver crashes at system startup
- Cleaned startup programs/DLLS with AutoRuns
- Disable Auto-grow for page file size, as this is limited to 1/8th of volume, and set it to 2 times the RAM size + 10 MB
- Disable Hibernation and thus Fast Startup, using "powercfg /hibernate off" or "powercfg -h off"
- Asus AI Suite disabled/re-enabled: disabled unwanted scheduled tasks for it, including disabling both Performance and DIP Away Mode and their "vCore Downgrade" settings in EPU
- Removed Armoury Crate with the uninstall tool: unrelated as issue also happened when it was never installed
- Reconfigured TDR in registry using "E:\Bin\TDR Manipulator.exe": values are now set to Enabled, 10, 9, 120 and 10 + apply and reboot --> did not fix it
- Disabled Brother Printer software autolaunching after startup
- For further troubleshooting only, use Process Monitor --> Options --> Enable Boot Logging to check what happens in background on startup
- Set in E:\Bin\nvidiaProfileInspector\nvidiaProfileInspector.exe - 5. Common; CUDA - Force P2 state on OFF --> now re-enabled as did not work
- Removed Shader Caches after closing all related applications. Disabled the option in the NVIDIA driver, rebooted and re-enabled it. Also cleaned %LocalAppdata%\D3DSCache and %LocalAppdata%\NVIDIA (both DXCache and GLCache folders) and Direct X-cache using cleanmgr
- MODS/MATS VRAM Test: all memory banks of current GPU are fine
- Replaced PSU: no fix
- Replaced GPU: did not fix it either

These actions are for troubleshooting and can each time be performed:
- Check folders C:\Windows\Temp\WER-XXX, C:\ProgramData\NVIDIA Corporation\CrashDumps, C:\Windows\Livekernelreports\Watchdog and C:\Programdata\Microsoft\Windows\WER for DMP-files + debug in WinDbg, to be sure the cause cannot be narrowed down to something else

These actions can still be done as a workaround:
- Set "Always 3D Clock" in GPU Tweak, however this disables 0dB Mode and consumes more power
- Set Power Management Mode to "Prefer Max Performance" in Nvidia Driver Control Panel - however, this disables 0dB Mode and consumes more power

These actions can still be performed:
- Flash this particular GPU with new vBIOS (there is a newer one for my card but can be tricky and I don't believe in hardware issue, so won't do that)
- Update mainboard BIOS (there is again a newer one for my mobo but I don't believe in hardware issue, so won't do that)
- Replace mainboard,memory,CPU,everything again! (but I don't believe in hardware issue, so won't do that)

I think about giving up to keep repeating the steps above, and rather start implementing workaround 1 now: Enabling "Always 3D Clock" in GPU Tweak, which is enabled at startup.

I got one. After updating to nvidia's V.531.41 for new RTX4090 and Mandatory Update for MSFS2020. Doesn't do it anywhere else.Just MSFS2020. Not in DCS or youtube. Always same spot and never had one in RTX3090.

View attachment 289398

This is certainly unrelated to this thread:

1. You have a RTX 40xx card
2. It's only in a game, not in desktop 2D mode
3. The particular game, FS 2020, is known for artifacting (even on Xbox there appear to be some glitches), due to temporal AA, nvidia settings/drivers and bugs in the game itself

Strange artifacts on top of screen (video)

As you can see in the video, every now and then when panning, out of nowhere, these artifacts will start appearing on the top part of my screen. I am not sure why it does it but itll do it out of nowhere, randomly and doesnt go away until I restart the sim. All my temps are fine, GPU around 65.

www.avsim.com

Mine's an ASUS TUF OC - bought launch day, so it's 100% the initial hardware revision if any came after that.

Here also a factory-OC'ed ASUS card. But I don't think hardware is related, as it happens with many type of GPU's, and also so rare.

Unstable VRAM, nah don't believe so, in that case it would happen at least once in a games too, and above all, the MODS/MATS results would have shown me.
F.i. the black artifacts which I saw today were parts of something which had appeared on the screen seconds before. In fact, I could visually still recognize parts of my CMD Window which launches a custom script at startup and which was still displayed from the memory cache.

That's why I still think something goes wrong between Windows and the Nvidia driver at startup, just can't explain why this does not happen all the time.

Video Memory Management and GPU Scheduling - Windows drivers

Video Memory Management and GPU Scheduling

learn.microsoft.com

For me, the cause still has to be located in one of the following phases:
Handling Memory Segments
Handling Command and DMA Buffers
GDI Hardware Acceleration
Video memory offer and reclaim
GPU preemption

Especially the last one seems to be highly potential as related (but not necessarily as cause, might also be consequence), at least for my particular problem. Quoted from that section:
If long-running packets cannot be successfully preempted, high-priority GPU work, such as work required by the Desktop Window Manager (DWM), can be delayed, resulting in glitches during window transitions and animations. Also, long-running GPU packets that cannot be preempted can cause a TDR process to repeatedly reset the GPU, and eventually a system bugcheck can occur.

As I have mentioned here before, when starting to tinker with GPU PreEmption using the QuickBoost software, the artifacts started to appear.

 

[Dr. Dro]

You make a good case for it being related to preemption, now I'm curious whether it's a driver or hardware bug?

 

[TheoneandonlyMrK]

Just had the artifacts too again, as usual after a cold startup this morning. Last time was on 6th of Dec 2022 with the previous driver version.
When I started recording them with Geforce Experience, they disappeared. Stopped the recording, they appeared seconds later again (on the recording also nothing special). Then some seconds later the screen became black and Windows 11 was started to get freezing, so I briefly pushed the power button and the pc succeeded just in time to turn off properly.

Again, in the event log as usual, first:
\Device\Video3
Graphics Exception: ILLEGAL_OPCODE

\Device\Video3
Graphics Exception: ESR 0x40a790=0x80000024

... and as a result of that, Windows TDR kicks in (you can see that this is only consequence, not cause, as it happens seconds later):
Display driver nvlddmkm stopped responding and has successfully recovered.

It started happening 5 minutes after booting up, and without opening any application/browser.
These actions were already performed one or many times, but did not fix the issue:
- Update GPU to latest driver after using DDU in safe mode
- Updated chipset to latest driver after using DDU in safe mode
- Updated mainboard to latest BIOS + loaded defaults
- Experimental Features of Geforce Experience disabled
- Hardware-accelerated GPU scheduling in Windows disabled/did not matter so afterwards re-enabled
- Disabled hardware acceleration individually when possible for all used applications
- iCue Enable Plugins disabled/re-enabled + deleted folders ASUS and NvidiaPlugin
- GPU Tweak 2/3 disabled/did not matter so afterwards re-enabled, with Gaming as default profile
- Set GPU in Line-Based Mode instead of default MSI using "E:\Bin\MSI_util_v3.exe" and rebooted/now reverted to MSI-mode
- Disabled/re-enabled Resizeable Bar in BIOS
- Moved GPU to another pci-e slot of the mainboard/Changed PCIe X16 slot Mode which should have options for Auto, Gen1, Gen2, Gen3 or Gen4. If its in Auto/Gen4, try to set it to Gen 3. If its in Gen 3, set try it to Auto
- Switch to other vBIOS when Dual vBios GPU, afterwards power off/reboot
- Disabled OC on CPU and RAM
- Memtest RAM: 0 errors
- Underclocked GPU
- Run GPU in Nvidia Control Panel Debug Mode (when supported) to reset it to reference clock speed: cannot be saved so no valid option for driver crashes at system startup
- Cleaned startup programs/DLLS with AutoRuns
- Disable Auto-grow for page file size, as this is limited to 1/8th of volume, and set it to 2 times the RAM size + 10 MB
- Disable Hibernation and thus Fast Startup, using "powercfg /hibernate off" or "powercfg -h off"
- Asus AI Suite disabled/re-enabled: disabled unwanted scheduled tasks for it, including disabling both Performance and DIP Away Mode and their "vCore Downgrade" settings in EPU
- Removed Armoury Crate with the uninstall tool: unrelated as issue also happened when it was never installed
- Reconfigured TDR in registry using "E:\Bin\TDR Manipulator.exe": values are now set to Enabled, 10, 9, 120 and 10 + apply and reboot --> did not fix it
- Disabled Brother Printer software autolaunching after startup
- For further troubleshooting only, use Process Monitor --> Options --> Enable Boot Logging to check what happens in background on startup
- Set in E:\Bin\nvidiaProfileInspector\nvidiaProfileInspector.exe - 5. Common; CUDA - Force P2 state on OFF --> now re-enabled as did not work
- Removed Shader Caches after closing all related applications. Disabled the option in the NVIDIA driver, rebooted and re-enabled it. Also cleaned %LocalAppdata%\D3DSCache and %LocalAppdata%\NVIDIA (both DXCache and GLCache folders) and Direct X-cache using cleanmgr
- MODS/MATS VRAM Test: all memory banks of current GPU are fine
- Replaced PSU: no fix
- Replaced GPU: did not fix it either

These actions are for troubleshooting and can each time be performed:
- Check folders C:\Windows\Temp\WER-XXX, C:\ProgramData\NVIDIA Corporation\CrashDumps, C:\Windows\Livekernelreports\Watchdog and C:\Programdata\Microsoft\Windows\WER for DMP-files + debug in WinDbg, to be sure the cause cannot be narrowed down to something else

These actions can still be done as a workaround:
- Set "Always 3D Clock" in GPU Tweak, however this disables 0dB Mode and consumes more power
- Set Power Management Mode to "Prefer Max Performance" in Nvidia Driver Control Panel - however, this disables 0dB Mode and consumes more power

These actions can still be performed:
- Flash this particular GPU with new vBIOS (there is a newer one for my card but can be tricky and I don't believe in hardware issue, so won't do that)
- Update mainboard BIOS (there is again a newer one for my mobo but I don't believe in hardware issue, so won't do that)
- Replace mainboard,memory,CPU,everything again! (but I don't believe in hardware issue, so won't do that)

I think about giving up to keep repeating the steps above, and rather start implementing workaround 1 now: Enabling "Always 3D Clock" in GPU Tweak, which is enabled at startup.

View attachment 289423


This is certainly unrelated to this thread:

1. You have a RTX 40xx card
2. It's only in a game, not in desktop 2D mode
3. The particular game, FS 2020, is known for artifacting (even on Xbox there appear to be some glitches), due to temporal AA, nvidia settings/drivers and bugs in the game itself

Strange artifacts on top of screen (video)

As you can see in the video, every now and then when panning, out of nowhere, these artifacts will start appearing on the top part of my screen. I am not sure why it does it but itll do it out of nowhere, randomly and doesnt go away until I restart the sim. All my temps are fine, GPU around 65.

www.avsim.com

Here also a factory-OC'ed ASUS card. But I don't think hardware is related, as it happens with many type of GPU's, and also so rare.

Unstable VRAM, nah don't believe so, in that case it would happen at least once in a games too, and above all, the MODS/MATS results would have shown me.
F.i. the black artifacts which I saw today were parts of something which had appeared on the screen seconds before. In fact, I could visually still recognize parts of my CMD Window which launches a custom script at startup and which was still displayed from the memory cache.

That's why I still think something goes wrong between Windows and the Nvidia driver at startup, just can't explain why this does not happen all the time.

Video Memory Management and GPU Scheduling - Windows drivers

Video Memory Management and GPU Scheduling

learn.microsoft.com

For me, the cause still has to be located in one of the following phases:
Handling Memory Segments
Handling Command and DMA Buffers
GDI Hardware Acceleration
Video memory offer and reclaim
GPU preemption

Especially the last one seems to be highly potential as related (but not necessarily as cause, might also be consequence), at least for my particular problem. Quoted from that section:
If long-running packets cannot be successfully preempted, high-priority GPU work, such as work required by the Desktop Window Manager (DWM), can be delayed, resulting in glitches during window transitions and animations. Also, long-running GPU packets that cannot be preempted can cause a TDR process to repeatedly reset the GPU, and eventually a system bugcheck can occur.

As I have mentioned here before, when starting to tinker with GPU PreEmption using the QuickBoost software, the artifacts started to appear.

Very thorough I am intrigued.

There are forum user's that think all these Issues are the user's fault.

I can't wait to hear they're explanation for why this is your fault ( I personally think your onto something)

 

[ThirtyNinety]

You make a good case for it being related to preemption, now I'm curious whether it's a driver or hardware bug?

For me, it smells too much like software/OS. Also, preemption is a good thing, it’s even needed within the current WDDM for videodrivers and one should not tinker with it.

So if this artifact issue really is related to it, what happens with the system at startup which prevents it from working as expected sometimes?

 

[Dr. Dro]

For me, it smells too much like software/OS. Also, preemption is a good thing, it’s even needed within the current WDDM for videodrivers and one should not tinker with it.

So if this artifact issue really is related to it, what happens with the system at startup which prevents it from working as expected sometimes?

I don't think it's the OS, or it'd manifest in a much broader range of hardware, remember, millions of people use Windows. It's gotta be drivers, or a hardware bug and it must be more specific to the GA102, and even then a bug of a more obscure nature indeed. What keeps intriguing me is that this only occurs in low power states and never otherwise.

 

[ThirtyNinety]

I don't think it's the OS, or it'd manifest in a much broader range of hardware, remember, millions of people use Windows. It's gotta be drivers, or a hardware bug and it must be more specific to the GA102, and even then a bug of a more obscure nature indeed. What keeps intriguing me is that this only occurs in low power states and never otherwise.

Indeed, but what I meant was "software in relation to the OS or to other software (like the Nvidia driver) or vice versa".
Remember this MSI Center story f.i. which caused artifacts with AMD drivers :

Occasional artifacts on system boot up - rx 570

Recently built my pc, had to buy a used GPU off someone on facebook for the build given the lack of options and prices in the last year or so. Anyway, since the build was completed I Get random driver issues/checkered screens (see photo below) which usually only happens right after I boot up. I...

community.amd.com

Related to the low power states: that is indeed what I will now try to figure out the upcoming days/weeks/months by forcing the GPU into P0 state each time my system boots (using GPU Tweak or Maximum performance doesn't matter actually).

If the problem never occurs again, it is 1:1 related to the P-state.

Just sorted the logs I gathered from yesterday's issue.
I've now come to the first conclusion that the issue just got worse by opening outlook after noticing the artifacts.
So when the artifacts are already there, using a hardware accelerated app like a browser or Outlook can make your system crash if you do not shutdown the pc in time.

I assume that hardware acceleration is NOT the cause. The request to hardware accelerate just cannot be handed off by the Windows GPU scheduler to the Nvidia Driver, because the latter is already having too much problems at that moment.
In the "video queue", there is some long running task active (what would it be? that's the BIGGEST question) which cannot be paused to switch context to the new priority task: display Outlook as hardware-accelerated. As a result, the OS decides after multiple retries (within seconds) to reseat the GPU driver/also known as TDR.

07:58:37 The operating system started.
07:58:42 Desktop Window Manager is now active.
07:58:44 Logon completed, artifacts are visible on the desktop.
08:01:36 Geforce Experience desktop recording launched, artifacts disappear temporarily.
‏‎08:01:42 Geforce Experience desktop recording stopped, artifacts reappear on exact the same spots.
08:02:19 Opening Outlook to see if hardware acceleration does make any difference. It does, artifacts are still visible but pc starts to get less responsive, Nvidia driver clearly begins to crash.
08:03:08 Graphics Exception: ILLEGAL_OPCODE is logged
08:03:08 Graphics Exception: ESR 0x40a790=0x80000024 is logged
08:03:08 Error occurred on GPUID: 800 is logged
08:03:09 Screen turns black and comes back again as a black screen with big blue/white rectangular artifacts. Display driver nvlddmkm stopped responding and has successfully recovered is logged. PC is almost unusable at this point.
08:03:44 Power Off (using the power button). PC succeeds in shutting down properly. Upon next cold boot, no issues can be noticed.

 

[ThirtyNinety]

Troubleshooting related to the P-states now enabled in an automated fashion using nvidia-smi.
On forehand you need to find the maximum (and/or minimum, if you don't want to reset it to defaults but always want to lock it to a certain speed within that range) core and memory clock values for your specific GPU.

So what I did is put this in a PowerShell script which runs in a scheduled task as SYSTEM on computer startup.
It seems to override the clocks, which the Nvidia driver always resets to the defaults, just perfectly.
And after two minutes, the script automatically resets both forced clock speeds back to the defaults, and thus the P-state goes back to P8 or whatever state it should have.

 

[NEC_HaoZ]

Troubleshooting related to the P-states now enabled in an automated fashion using nvidia-smi.
On forehand you need to find the maximum (and/or minimum, if you don't want to reset it to defaults but always want to lock it to a certain speed within that range) core and memory clock values for your specific GPU.

So what I did is put this in a PowerShell script which runs in a scheduled task as SYSTEM on computer startup.
It seems to override the clocks, which the Nvidia driver always resets to the defaults, just perfectly.
And after two minutes, the script automatically resets both forced clock speeds back to the defaults, and thus the P-state goes back to P8 or whatever state it should have.

View attachment 289567

Thanks for your work in finding the cause of the issue.
On Chinese forums, I have seen that many 30 series notebooks also have this issue, which usually appears on the RTX3060 notebook version.
Notebook manufacturers do a lot of compatibility testing for their products, so the probability of the problem occurring in the combination of BIOS and graphics cards is relatively small, but this problem also exists. Therefore, I think the problem is the same as you described, The reason is more likely to be the compatibility of NV drivers and Windows.

 

[NEC_HaoZ]

So I actually created a similar issue with mine today. I did a full shutdown and booted back into the BIOS. Did a case reset, loaded up the default values (no DRAM profiles) and rebooted. Then directly back into the bios. Loaded my old profiles but made one change which was disabling the Serial Com port. Rebooted into Windows 11 normal but then suddenly got the rectangle artifacts. Then the PC just rebooted on its own, no BSOD. I updated the NVidia driver on my RTX 3060 and rebooted. The problem hasn't returned. I've never seen this before but it only occurred after I went into the BIOS and did the things I mentioned previously. Also never had this PC just reboot on its own with no Windows warning or BSOD. Somewhere there was a hardware/Windows conflict when it rebooted into Windows the first time. And BTW, I left that serial port disabled so that wasn't it. I also don't think it was a graphics driver problem.

I think these motherboard BIOS's become unstable from time to time. I did a full case reset and BIOS default/reboot thing because it always fixes a problem with the on-board Realtek NIC occasionally not loading up after the system comes out of sleep/hibernation. A full shut down with the power supply button off along with a BIOS case reset is the only thing that fixes that issue. This time, I did a couple additional things and created this stability problem at the motherboard level, which affected Windows in two ways. One with the rectangles and two, with the sudden reboot. If I see this again, I'm going to swap out the DRAM modules and see if maybe that's the source of my instability. Maybe the DRAM stick's EMP's are being affected over time and especially when the PC comes out of sleep mode.

Not long after the issue occurred first time.I was try to replaced DRAM and it has no effect.

 

[alutyl]

Thanks for your work in finding the cause of the issue.
On Chinese forums, I have seen that many 30 series notebooks also have this issue, which usually appears on the RTX3060 notebook version.
Notebook manufacturers do a lot of compatibility testing for their products, so the probability of the problem occurring in the combination of BIOS and graphics cards is relatively small, but this problem also exists. Therefore, I think the problem is the same as you described, The reason is more likely to be the compatibility of NV drivers and Windows.

I've seen posts on reddit that people have installed a new version of the bios on their laptop, and after that these artifacts have completely disappeared.

 

[Zer0code]

Guys, can someone write to nvidia technical support about our problem? Send them the link to this thread, send them sample videos and screenshots. Write to tech support specifically, not the nvidia forum! I can't do it because my English is not perfect yet. But you can!

 

[alutyl]

Guys, can someone write to nvidia technical support about our problem? Send them the link to this thread, send them sample videos and screenshots. Write to tech support specifically, not the nvidia forum! I can't do it because my English is not perfect yet. But you can!

I already wrote to nvidia support. They told me to install the latest version of the drivers. I described the problem in great detail. I sent a lot of screenshots of different users with our problem. They told me that NO ONE had contacted them with this problem before me. I asked several times what I could do so they could fix the problem but they told me to just install a new driver and the problem should go away. They are not going to do anything about this problem if you look at the way nvidia support answered me.

 

[Waldorf]

artifacts are 99.9% of the time related to clocks above whats stable, or not enough juice.
there were enough times in the past where changes (hw) caused it, even that ppl didnt oc.
and what "you" believe is kind of irrelevant, if you havent tried it (e.g. bios update), dont say it wont fix it.

e.g. switching from samsung to hynix ram, with the hynix unable to ramp up clocks as fast (as the samsung),
and thus needed more voltage (bios fix), or downclocking vram a little (less than 5Mhz), to eliminate the problem,
so anytime a rig has trouble with artifacts, i do test with lower vram clocks (afterburner). and turn off hw acceleration in things like browsers,
another try would be setting pcie to 3.0 (and 8x).

just cant see why ppl (with problems) still seem to refuse to even try something like that (underclock).
who cares if you "didn't oc", as long as there's a chance it will fix the issue.

 

[ThirtyNinety]

artifacts are 99.9% of the time related to clocks above whats stable, or not enough juice.
there were enough times in the past where changes (hw) caused it, even that ppl didnt oc.
and what "you" believe is kind of irrelevant, if you havent tried it (e.g. bios update), dont say it wont fix it.

e.g. switching from samsung to hynix ram, with the hynix unable to ramp up clocks as fast (as the samsung),
and thus needed more voltage (bios fix), or downclocking vram a little (less than 5Mhz), to eliminate the problem,
so anytime a rig has trouble with artifacts, i do test with lower vram clocks (afterburner). and turn off hw acceleration in things like browsers,
another try would be setting pcie to 3.0 (and 8x).

just cant see why ppl (with problems) still seem to refuse to even try something like that (underclock).
who cares if you "didn't oc", as long as there's a chance it will fix the issue.

To summarize this thread for you, this is not the "typical artifact problem because of gpu" which you seem to describe here, at least it is not in my case.
As the weird thing is:
regardless which nvidia GPU I put in this pc, I never have artifacts, only once every 3-6 months and in that case immediately at Windows startup, always accompanied with a nvidia driver crash and immediately fixed if I reboot.

So I try to reproduce (at least if these are related to high P-state, afterwards also need to do the same as well for low P-state) these artifacts now by launching an OC (without using any third-party tool, just by manipulating the nvidia driver itself) of the gpu immediately on Windows startup to the values maximally supported by the hardware, and not even 1 additional MHz.
I could permanently underclock and maybe this will fix it forever, and maybe it won’t. But for me that is the same as disabling hardware acceleration in your programs:
avoiding, instead of solving a problem with a piece of hardware for which you paid a lot of money.

And next to that, I'd also like to know the root cause or at least the conditions in which this particular issue manifests itself

 

[KsushaTeaKisa]

I created a separate thread about problems with artifacts like "flickering black checkered squares" due to the fact that in this thread there are two different problems with artifacts and it need to be discussed separately, because they are most likely not related to each other. I'm talking about artifacts like these users have: Zer0code, Tarkovski and alutyl in this thread.
I talked about every detail I could understand about this issue, including partial solutions. So please check my topic.

 

[Waldorf]

@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.