• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

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

No matter how you spin this I can't see how it can be related to drivers/vBIOS, those look 100% like memory artifacts.
 
No matter how you spin this I can't see how it can be related to drivers/vBIOS, those look 100% like memory artifacts.
This thread has already made it very clear that not everyone knows what a driver graphical corruption is.
 
No matter how you spin this I can't see how it can be related to drivers/vBIOS, those look 100% like memory artifacts.
[GeForce RTX 4090] Watch Dogs 2 may display flickering when staring at the sky
Players report black/grey screens in Outer Wilds with 522.25 driver [3841593]
Lumion Pro 12.3 - Heavy corruption observed on app window [3784371]
Tiny Tina's Wonderlands displays texture corruption after extended gameplay on NVIDIA GPUs [3777340]
[Forza Horizon 5] Rainbow like artifacts in game after driver update. [3685123]
[Call of Duty: Modern Warfare II] Flashing corruption can be seen randomly while playing the game. [3829010]
[Chivalry 2] Toggling DLSS preset may cause gameplay to shimmer or show black rectangle. [3692024]
[Prepar3D] Light sources display flashing black boxes.
[Call of Duty: Vanguard]: The game may display random corruption. [3503111]
[Battlefield 2042]: The game may display color flashes on the screen. [3503086]

I can go more if you want ... those are just random artifacts/graphic corruption picked from nvidia drivers notes.

Now tell me that the same graphic corruption are in thise notes, it cannot be in the desktop because hey, is has to be GPU memory faulty !!!! Nevermind I had on few different cards, naah that dosent matter...
 
[GeForce RTX 4090] Watch Dogs 2 may display flickering when staring at the sky
Players report black/grey screens in Outer Wilds with 522.25 driver [3841593]
Lumion Pro 12.3 - Heavy corruption observed on app window [3784371]
Tiny Tina's Wonderlands displays texture corruption after extended gameplay on NVIDIA GPUs [3777340]
[Forza Horizon 5] Rainbow like artifacts in game after driver update. [3685123]
[Call of Duty: Modern Warfare II] Flashing corruption can be seen randomly while playing the game. [3829010]
[Chivalry 2] Toggling DLSS preset may cause gameplay to shimmer or show black rectangle. [3692024]
[Prepar3D] Light sources display flashing black boxes.
[Call of Duty: Vanguard]: The game may display random corruption. [3503111]
[Battlefield 2042]: The game may display color flashes on the screen. [3503086]

I can go more if you want ... those are just random artifacts/graphic corruption picked from nvidia drivers notes.

Now tell me that the same graphic corruption are in thise notes, it cannot be in the desktop because hey, is has to be GPU memory faulty !!!! Nevermind I had on few different cards, naah that dosent matter...

Flickering/texture corruption and whatever else is listed there does not look like what is shown in this thread which looks like classic memory artifacting. Do you not find it odd that this hasn't been fixed yet ? 3000 series is like 3 years old at this point, doesn't that ring a bell that this might actually be hardware ?

Anyway, if you still think it's software that's fine, I am just wondering what exactly is the resolution you seek to achieve here, because as I pointed out these cards have been out for quite some time and if this was fixable it should have been dealt with long ago.
 
Had a similar issue with my 3090, thought the card was dying cause i mined heavily on it and did some crazy overclocks with 600 watt power limits.

I was randomly seeing artifacts and sometimes the screen kept flickering every 20 seconds, only restarting would fix it.

Turned out that it was the combination of hdmi and hdmi audio drivers that caused it. I installed older drivers and the card works just fine with no issues for the last 3 months.
 
Flickering/texture corruption and whatever else is listed there does not look like what is shown in this thread which looks like classic memory artifacting. Do you not find it odd that this hasn't been fixed yet ? 3000 series is like 3 years old at this point, doesn't that ring a bell that this might actually be hardware ?

Anyway, if you still think it's software that's fine, I am just wondering what exactly is the resolution you seek to achieve here, because as I pointed out these cards have been out for quite some time and if this was fixable it should have been dealt with long ago.

Don't get the logic here why the age of a video card series would be a relevant argument here to conclude that these artifacts are certainly related to a hardware issue, as these only happen upon a driver crash after Windows startup.

I can recall, without effort, many cases where similar graphical corruption was introduced without any hardware being defect. It even happened on my NUC last year, a well known issue with a specific driver which was caused by Windows Update.

"if this was fixable it should have been dealt with long ago." --> that just puts too much faith in a money greedy company like Nvidia imo.
 
Huh. Different cards. If your thermals are good to go... how's your system RAM? Not VRAM. I had similar issues with an old system (intermittent black squares, drivers restarting, occasional rainbow bs) with an R9 290, and a GTX 1070 I replaced it with. System restart, and the problem would go away for a time.


End result -RAM sticks were bad. After a few years of running at advertised speeds, they crapped out. Started with graphical glitches here and there. Graphical glitches there often. Several months later ended with random blue screens of death. They (both sticks) failed MemTests at stock, and XMP settings. Adjusting voltage, and timings didn't help. Replaced RAM. No more issues. R9 290, and GTX 1070 were fine.
 
I can recall, without effort, many cases where similar graphical corruption was introduced without any hardware being defect.

Similar does not mean the same, memory artifacting has a very distinct look, anyway, good luck with this.
 
Apparently 1 month ago, someone was able to reproduce the issue by performing some Nvidia Tweaks from the QuickBoost app on Github for the installed driver.
To test with on your own risk of course.


Url for example purposes, not necessarily triggering the issue because I did not test it myself yet:
 
Last edited:
This morning I turned my PC on and before I could even sit on my chair the desktop had square artifacts.
I clicked on Stellaris shortcut and before it could even open the game launcher the monitor lost signal and the PC restarted, didn't even see a blue screen.

If this issue only happens during cold boots and once every 1-2 months then I'm going to try to RMA in 2,5 years when the warranty almost ends and maybe get a better GPU for free.
 
Last edited:
The only good answer here is an RMA, regardless of what causes it. The card is not performing as advertised - it being hyper sensitive to the boot process is just not a good thing and not a you-problem either. To each their responsibilities, right?
 
The only good answer here is an RMA, regardless of what causes it. The card is not performing as advertised - it being hyper sensitive to the boot process is just not a good thing and not a you-problem either. To each their responsibilities, right?
Agree, but still, what if you RMA'ed three times already like OP, and still have the same issue with different GPU's and builds?

In that case, RMA is actually more pointless then not RMA'ing. As in the mean time you can keep your GPU, just reboot when it occurs and figure out for yourself how to make it working somehow, until it's already time for a next gen gpu, which is hopefully better ...

It has already been 6 weeks now, and still haven't seen the issue back even once, fingers stay crossed ...

This morning I turned my PC on and before I could even sit on my chair the desktop had square artifacts.
I clicked on Stellaris shortcut and before it could even open the game launcher the monitor lost signal and the PC restarted, didn't even see a blue screen.

If this issue only happens during cold boots and once every 1-2 months then I'm going to try to RMA in 2,5 years when the warranty almost ends and maybe get a better GPU for free.
A blue screen is just not always visible. You should certainly check the relevant folder c:\Windows\LiveKernelReports\WATCHDOG or the reliability history viewer in Windows, or check for system event logs with source nvlddmkm event ID 4101 and 13 as well.
 
Last edited:
Agree, but still, what if you RMA'ed three times already like OP, and still have the same issue with different GPU's and builds?

In that case, RMA is actually more pointless then not RMA'ing. As in the mean time you can keep your GPU, just reboot when it occurs and figure out for yourself how to make it working somehow, until it's already time for a next gen gpu, which is hopefully better ...

It has already been 6 weeks now, and still haven't seen the issue back even once, fingers stay crossed ...


A blue screen is just not always visible. You should certainly check the relevant folder c:\Windows\LiveKernelReports\WATCHDOG or the reliability history viewer in Windows, or check for system event logs with source nvlddmkm event ID 4101 and 13 as well.
If the problem persists you can cancel the contract as well. Product non conformity, simple as. Whether driver or OS related... companies need to get their shit together. Of course your path is valid too, but in the simple consumer sense, I wouldnt accept this. Give me back my money or replace with a model/tier that does not suffer the problems and has equal or better perf.

Fact is, if RMA is accepted you have a strong case. Who's to say the company isnt just juggling around faulty GPUs? What certainties should you have as a paying customer?
 
Last edited:
Old-film-like artifacts=Unstable VRAM most likely. (Same with "Cache Hierarchy Error" WHEA)

Squares and random Direct3D init failed error message while already initialized= Unstable GPU core most likely.

Skipped frames=VRAM most likely!

Similar does not mean the same, memory artifacting has a very distinct look, anyway, good luck with this.
Unstable core (GPU) has a very raster-y mess, typically, also driver restarts tend to lean towards core, even when I saw that logged before for unstable VRAM on my GTX 1660 Super, after a random Halo hang, where Gearbox Halo CE is dumb sometimes, it thought I lost server connection, LOL! At least its UI can sure be dumb AF! I've possibly seen incorrect messages by its UI before.

You can think of unstable core as looking like a bad digital TV signal!

Unstable VRAM, OTOH, can look just like an analog TV mess.
 
Last edited:
If the problem persists you can cancel the contract as well. Product non conformity, simple as. Whether driver or OS related... companies need to get their shit together. Of course your path is valid too, but in the simple consumer sense, I wouldnt accept this. Give me back my money or replace with a model/tier that does not suffer the problems and has equal or better perf.

Fact is, if RMA is accepted you have a strong case. Who's to say the company isnt just juggling around faulty GPUs? What certainties should you have as a paying customer?
That is also true, but even additional reason why I did not RMA my gpu when this issue started to happen. Instead I did intensive checks for every memory bank of the VRAM with MODS/MATS and also a lot of core stability tests for GA102 and the results assured me this gpu is perfectly fine.
 
Last edited:
Had a similar issue with my 3090, thought the card was dying cause i mined heavily on it and did some crazy overclocks with 600 watt power limits.

I was randomly seeing artifacts and sometimes the screen kept flickering every 20 seconds, only restarting would fix it.

Turned out that it was the combination of hdmi and hdmi audio drivers that caused it. I installed older drivers and the card works just fine with no issues for the last 3 months.
I suspect a bad HDMI signal causes symptoms resembling core OC'ed too much! (GPU)
 
I suspect a bad HDMI signal causes symptoms resembling core OC'ed too much! (GPU)
But doesn't cause these intermittent nvidia driver crashes.
As these are a mandatory key indicator for this particular issue described in this thread ...
 
But doesn't cause these intermittent nvidia driver crashes.
For me, it looked like it caused random driver restarts, which is the meme of getting too greedy with the core!
 
Yeah it’s kinda sad, for me this is clearly a software-related one, specifically linked to what happens during Windows startup.

But I must say, it has been now one month since I saw the latest driver crash at Windows startup together with these small artifacts, whereas before I started to get the issue almost weekly.
Yes, my GPU model is 3080 12G. After the small square issue, I have replaced two new GPUs of different brands, but they have the same problem.
As you described, it is related to the behavior of NV driver at startup. And it only happens in 2D mode. In China's forums, I have seen a lot of the same issue.
Usually, people will think that this is the reason for VRAM, but in fact it is a bug with the driver. It may be the unreasonable configuration of VRAM or GPU core by the driver or GPU firmware at startup.
However, NV will not solve this issue because they are a large company and the efficiency of solving the issue is very low.
 
I had it too. And I was so sad until I saw this thread.
 
I had it too. And I was so sad until I saw this thread.
Hi,
Yep I forgot all about the nvidia 30 series driver brain fart issue.
 
ASUS TUF RTX 3060 GAMING OC, Having same issue but everything running fine.

Thanks for the thread. I figured it was just a bug. I see it happen quite often when I'm streaming discord and switching between windows.
 
I tried to edit the first post but I cant, so I will paste here more videos related to the issue itself



 
If it's a hardware problem, why the problem resolves after restart? And why there is no glitch in the bios screen?
 
Back
Top