Question Random artifacts with RTX 3080

[Osquiworld]

I have been experiencing seemingly random artifacting with my Gigabyte RTX 3080 for at least a year and a half. The screen will get covered with green and purple squares and will either freeze forcing me to hard reboot or after a few seconds it will go black and come back to desktop. I have already tried:

-Uninstalling drivers with DDU and reinstalling them
-Changing both the cables and the PCIe slot the GPU is seated on
-Cleaning dust off (multiple times)
-Underclocking both the core and memory clock with Afterburner

None of these things have helped. The artifacts still appear from time to time, and some times it will crash 10+ times a day for a week and then spend another whole week with no crashes. I don't think it's an overheating problem as it happens even when the GPU isn't under any load, just on desktop. I want to make sure it's the GPU what's failing before possibly replacing it, but I have no way to reproduce the artifacts and no other system to try it on. I've read that this is usually a VRAM issue, but underclocking did not help, as I mentioned. Any advice on what can be causing it and if there's a fix would be really helpful.

I have a z490e motherboard, a 10700k CPU and a RM850 PSU, by the way.

https://imgur.com/a/N9BkgPt

 

[beyondlogic]

I have been experiencing seemingly random artifacting with my Gigabyte RTX 3080 for at least a year and a half. The screen will get covered with green and purple squares and will either freeze forcing me to hard reboot or after a few seconds it will go black and come back to desktop. I have already tried:

-Uninstalling drivers with DDU and reinstalling them
-Changing both the cables and the PCIe slot the GPU is seated on
-Cleaning dust off (multiple times)
-Underclocking the both the core and memory clock with Afterburner

None of these things have helped. The artifacts still appear from time to time, and some times it will crash 10+ times a day for a week and then spend another whole week with no crashes. I don't think it's an overheating problem as it happens even when the GPU isn't under any load, just on desktop. I want to make sure it's the GPU what's failing before possibly replacing it, but I have no way to reproduce the artifacts and no other system to try it on. I've read that this is usually a VRAM issue, but underclocking did not help, as I mentioned. Any advice on what can be causing it and if there's a fix would be really helpful.

I have a z490e motherboard, a 10700k CPU and a RM850 PSU, by the way.

https://imgur.com/a/N9BkgPt

If GPU is within warranty rma it that's a big technical issue that may be damaged vram chips or the GPU core itself.

If it's out of warranty

You can dismantle the GPU and replace the thermal pads and paste

Things required
Screw driver

Isopropanol alcohol
Cotton swabs ( not balls )


Thermal pads. Thermal grizzly minus.
For thickness contact gigabyte with your make and model they should be able to tell you.

Other thing to note a few of gigabytes cards started cracking at the PCB near the bottom of the card

View: https://youtu.be/0RO5rhrELFs?feature=shared

 

[Osquiworld]

If GPU is within warranty rma it that's a big technical issue that may be damaged vram chips or the GPU core itself.

Unfortunately, it is out of warranty.

You can dismantle the GPU and replace the thermal pads and paste

So, it's 100% a faulty GPU what is causing the artifacts? Because if that's the case I think I'd rather upgrade than try to open the card, I'm just worried that if I buy another card the artifacts will still be there. Also, does that mean that overheating may be causing the issue? Temps seem to be just fine, even under stressful conditions the artifacts only happen every once in a while, same as when there's barely any load.

Other thing to note a few of gigabytes cards started cracking at the PCB near the bottom of the card

I've checked and it doesn't appear to be bent or cracked, so I think it's safe to rule out that possibility.

 

[Phaaze88]

Also, does that mean that overheating may be causing the issue?

Gpu core?
Hot spot?
Memory junction?

 

[Osquiworld]

Gpu core?
Hot spot?
Memory junction?

I am a total noob regarding temperatures (and hardware in general), so I don't know what to make of these records. This is with only hwinfo open:

I've noticed that the fans are at 0%. Is that normal? They're set to auto on afterburner.
EDIT: with fans at 100% speed:

 

[Silas Sanchez]

Those temps at idle (~24C ambient) are fine. As a general rule under heavy loads you should keep gpu temp to 70-75C and hot spot to 85C. Going beyond that may start to be a problem longevity, and you really need to take into account case setup and ambient temp.

One of the big mysteries with graphics cards is how long do they last, and why exactly do they die. Could be dozens of factors like poor quality control.
In my experience when it artifacts the card is gone and is accompanied with sustained high temps and long hours. Welcome to the throw away society.

 

[Phaaze88]

I am a total noob regarding temperatures (and hardware in general), so I don't know what to make of these records. This is with only hwinfo open:

I've noticed that the fans are at 0%. Is that normal? They're set to auto on afterburner.
EDIT: with fans at 100% speed:

We need to see those thermals with the card under load.

Some gpu models have a 0rpm mode when the gpu core is under a preset target - I think it varies between vendors.

 

[Osquiworld]

We need to see those thermals with the card under load.

Some gpu models have a 0rpm mode when the gpu core is under a preset target - I think it varies between vendors.

This is running Forza Horizon 5 at 1440p, max settings, no DLSS:

But as I said earlier, the artifacts happen even when idle
EDIT: it crashed again twice in a span of 20 seconds right after posting this, with the game still running. As usual, it closed the game and went back to desktop. I don't know if this is relevant, but many times when it happens the windows device connect sound plays

 

[Phaaze88]

This is running Forza Horizon 5 at 1440p, max settings, no DLSS:

But as I said earlier, the artifacts happen even when idle
EDIT: it crashed again twice in a span of 20 seconds right after posting this, with the game still running. As usual, it closed the game and went back to desktop. I don't know if this is relevant, but many times when it happens the windows device connect sound plays

Evidence so far points to memory failing.
The memory junction is toasty. I wonder for how long it's been operating that high while you've had it - no way to know.
I know crash to desktops are usually driver/software related, but when audio plays or can be heard, it's also memory.

Do you run your browser with gpu hardware acceleration?

 

[Osquiworld]

Do you run your browser with gpu hardware acceleration?

I used to, until last summer. I also disabled it on Discord a few days ago.

 

[Phaaze88]

I used to, until last summer. I also disabled it on Discord a few days ago.

Ok. I was hoping that would've been the cause of the idle symptoms, but nope.

There's not much more to try - at least on your own. There's one more thing that comes to mind besides memory failure, that some of the solder joints connecting the gpu core to the PCB has separated or come loose from the weight of the card + heat cycles.

Prepare for a new card...

 

[Osquiworld]

Ok. I was hoping that would've been the cause of the idle symptoms, but nope.

There's not much more to try - at least on your own. There's one more thing that comes to mind besides memory failure, that some of the solder joints connecting the gpu core to the PCB has separated or come loose from the weight of the card + heat cycles.

Prepare for a new card...

I already knew that this was the most likely scenario, but it still hurts. I'll keep this one for now until I see a good deal. Thanks for all your help though! Really appreciate it.

 

[beyondlogic]

Unfortunately, it is out of warranty.

So, it's 100% a faulty GPU what is causing the artifacts? Because if that's the case I think I'd rather upgrade than try to open the card, I'm just worried that if I buy another card the artifacts will still be there. Also, does that mean that overheating may be causing the issue? Temps seem to be just fine, even under stressful conditions the artifacts only happen every once in a while, same as when there's barely any load.

I've checked and it doesn't appear to be bent or cracked, so I think it's safe to rule out that possibility.

This Is usually memory related on the GPU usually due to failed thermal pads. Usually takes a card to be brute forced for a long time to see that kind of damage so I would be willing to bet that either the thermal pads where not placed properly or they cheapest out on the thermal pads seen cards with like half a thermal pad on the memory before.

I've regularly opened up cards over the years and found many that the manufacturer forgot to put on.