RTX 3060 black screens on load and light browsing even after RMA

[Finix23]

Hi all, I recently bought a new rtx 3060 which worked fine for the first few days, but then started giving me black screens. Over the span of 2 weeks the frequency of black screens increased until it stopped giving me any screen.
So I sent the gpu in for warranty repair, and when I got it back the gpu now gives me a screen again, but black screens about 10-30 seconds into furmark or after opening a game, and sometimes also on very light load, like browsing or youtube.
When I got my gpu back, they also sent a screenshot of their furmark benchmark, and the gpu was being stressed for 30 minutes at high settings, so I doubt the gpu is still faulty.
I doubt the gpu is overheating since the temperatures are below 60° before it crashes.
Also, I tried another gpu (gtx 1650) and it works fine with no issues.
The event viewer shows "Display driver nvlddmkm has stopped working and successfully recovered" with error 4101 when it black screens.
Also, my PCIe lanes are linked to x8 instead of x16. I think this is due to my CPU, but I've read that this shouldn't be an issue but I could be wrong.

My specs:

• CPU: Ryzen 5 3400G
• RAM: G.Skill 32gb ram ddr4
• GPU: RTX 3060
• MOBO: AsRock B450 Pro4
• PSU: Seasonic Focus GX650

Both the psu and ram are new, I bought them less than 3 weeks ago. While I've had my cpu and motherboard for about 3 years.
I never did any overclocking if that matters.

I already tried:

• Performing a clean re-installation of drivers with DDU
• Reseating cpu and ram
• Reset BIOS CMOS
• Updated BIOS to latest version
• Connected HDMI instead of DP and tried other DP ports
• Mounted gpu in second PCIe slot (2.0 x8)
• Tried using an older nvidia driver (528.49)
• Disabled fast boot & windows auto-updating drivers
• Set power settings to "maximum performance" in nvidia control panel
• Tried enabling debug mode in nvidia control panel
• Reinstalled windows 10 22h2 (I updated very recently, I was on 21h2 and the issue already existed)
• Underclocking slightly with MSI afterburner

Also, the system does not crash after black screening, so no minidumps are created.

 

[thesmokingman]

I'm guessing you're running all stock right? And when it black screens, you wrote that it hasn't actually crashed? Check your monitor, unplug the power and replug and see if it comes back? In some cases the monitor can drop out. Also if you have another computer in the place install VNC on your comp and the spare and see if you can remote into yours to confirm the system state behind the black screen.

 

[Finix23]

I'm guessing you're running all stock right? And when it black screens, you wrote that it hasn't actually crashed? Check your monitor, unplug the power and replug and see if it comes back? In some cases the monitor can drop out. Also if you have another computer in the place install VNC on your comp and the spare and see if you can remote into yours to confirm the system state behind the black screen.

Yes, I'm running all stock. I forgot to mention I have a dual monitor setup, they both black screen and I've tried replugging them as you said. And I do not have another computer, but I can still hear sounds playing after the black screen and I can also reload the drivers with windows + control + shift + b and hear the beep.

 

[thesmokingman]

Are you running surround?

 

[Finix23]

Are you running surround?

No I'm just using headphones

But I also wanted to add that I tried connecting one monitor to the integrated graphics and the other to the gpu, and the result was that only the monitor connected to the gpu black screened with no signal and the other one kept functioning normally with no problems

And usually after I restart my pc after it black screens it takes an unusual long time for it to POST.

 

[thesmokingman]

No I'm just using headphones

But I also wanted to add that I tried connecting one monitor to the integrated graphics and the other to the gpu, and the result was that only the monitor connected to the gpu black screened with no signal and the other one kept functioning normally with no problems

And usually after I restart my pc after it black screens it takes an unusual long time for it to POST.

I'm referring to NVSurround. For now it reads like you have a config issue either with pcie slot priority, or monitor refresh? It's a bit vague right now but it feels like something to do with your panels and their config. Also I would not use black screening, that refers to something else.

 

[Finix23]

I'm referring to NVSurround. For now it reads like you have a config issue either with pcie slot priority, or monitor refresh? It's a bit vague right now but it feels like something to do with your panels and their config. Also I would not use black screening, that refers to something else.

I've never heard of NVSurround so I don't think I'm using it unless its something included by default
I don't know if it could be pcie slot priority or monitor refresh since the other gpu I tried works fine

Edit: Yes, NVSurround seems to be disabled. Sorry if it's not in English.

 

[thesmokingman]

I've never heard of NVSurround so I don't think I'm using it unless its something included by default
I don't know if it could be pcie slot priority or monitor refresh since the other gpu I tried works fine

Edit: Yes, NVSurround seems to be disabled. Sorry if it's not in English.

View attachment 315513

Ah, you have mismatched panels, that will generally give poor results if the panels are not matched.

 

[Finix23]

Ah, you have mismatched panels, that will generally give poor results if the panels are not matched.

Sorry I don't get what you mean, how are they mismatched?

 

[OneMoar]

open a RMA with the manufactor: any shop that uses 30m of furmark to test a card is not one you should be dealing with

 

[Finix23]

open a RMA with the manufactor: any shop that uses 30m of furmark to test a card is not one you should be dealing with

Alright, I'll open a RMA with the manufacturer. Thanks.

 

[thesmokingman]

Sorry I don't get what you mean, how are they mismatched?

You should research NVSurround. Using any gpu with mismatched panels is not ideal due size, refresh rate, resolution etc etc, ie. they are not the same monitors. By default ONLY one panel is active in gaming, unless you enable surround which creates a single panel out of the two matching panels which would then allow you to run a larger resolution.

Your OP is actually vague, cuz we don't know what screen is going dark, both or one?

 

[Finix23]

You should research NVSurround. Using any gpu with mismatched panels is not ideal due size, refresh rate, resolution etc etc, ie. they are not the same monitors. By default ONLY one panel is active in gaming, unless you enable surround which creates a single panel out of the two matching panels which would then allow you to run a larger resolution.

Your OP is actually vague, cuz we don't know what screen is going dark, both or one?

Both screens go dark, I also tried connecting only one monitor and the same issue happens

 

[thesmokingman]

Both screens go dark, I also tried connecting only one monitor and the same issue happens

It's not looking good so far. Try running a game in windowed mode and see what happens.

 

[Finix23]

It's not looking good so far. Try running a game in windowed mode and see what happens.

I've tried but it still black screens. By the way, it doesn't happen only when playing games, but also when browsing or watching youtube after like 1 or 2 hours.

 

[ty_ger]

Thos is usually a power supply issue, monitor cable issue, video card issue, or memory issue.

I would run memtest on your system RAM to verify stability, and try AMD's maximum CPU rated RAM speed if less than the memory manufacturer's rated RAM speed.

 

[Finix23]

Thos is usually a power supply issue, monitor cable issue, video card issue, or memory issue.

I would run memtest on your system RAM to verify stability, and try AMD's maximum CPU rated RAM speed if less than the memory manufacturer's rated RAM speed.

I've replaced power supply, cable and ram and the problem persists. Also my old gpu works fine in my system, so I'm pretty sure its the new gpu at this point.

 

[ty_ger]

Yes, I realize that you replaced those items. That's why I only suggested one additional thing.

I would verify that the system RAM is stable and test it at AMD's maximum suggested RAM speed for that CPU. It is surprisingly common for a newer GPU to have issues related to RAM stability when an older GPU doesn't. Freakishly common. I would test to be sure.

 

[TechHalp]

I would suggest trying a game with 1 screen unplugged, also check to make sure the GPU fan/s are running, and the temps on the gpu just before crashing. Try switching the power cable to the GPU as well, make sure you push it in until you feel the click

 

[eidairaman1]

Time to scrap the 3060

 

[Finix23]

Yes, I realize that you replaced those items. That's why I only suggested one additional thing.

I would verify that the system RAM is stable and test it at AMD's maximum suggested RAM speed for that CPU. It is surprisingly common for a newer GPU to have issues related to RAM stability when an older GPU doesn't. Freakishly common. I would test to be sure.

Tried putting RAM speed to AMD's maximum suggested of 2933 Mhz and the issue still persists.

I would suggest trying a game with 1 screen unplugged, also check to make sure the GPU fan/s are running, and the temps on the gpu just before crashing. Try switching the power cable to the GPU as well, make sure you push it in until you feel the click

I've tried only one monitor with HDMI and also tried another power cable. The GPU fans are still running at the same speed before it crashed (unlike other posts that said their fans would spin to maximum speed) and the temperature is around ~60 degrees when it crashes, but sometimes even lower (~40-50 degrees) since it also happens on idle.

 

[TechHalp]

Tried putting RAM speed to AMD's maximum suggested of 2933 Mhz and the issue still persists.


I've tried only one monitor with HDMI and also tried another power cable. The GPU fans are still running at the same speed before it crashed (unlike other posts that said their fans would spin to maximum speed) and the temperature is around ~60 degrees when it crashes, but sometimes even lower (~40-50 degrees) since it also happens on idle.

Very strange, would RMA again