Question 2080 Ti Founders Edition - - - artifacts & black screens after a bit of heavy use ?

[1vk3]

Hello.

SPECS:
GPU: NVIDIA RTX 2080 Ti Founders Edition
CPU: Ryzen 5 5600x
MOBO: ROG STRIX B450-F GAMING
PSU: EVGA Super Nova 650 G5, 80 Plus Gold 650W

I can't really tell exactly what's the problem with my 2080ti or my system. I recently bought this card second hand, and at first it worked great, for a couple days or so. I then started to encounter I guess driver resets (black screen, then back to windows, with whatever application I was using notifying me that my 3d device has disconnected), black screens to blue screens (most common errors being VIDEO TDR FAILURE, VIDEO SCHEDULER INTERNAL ERROR), and then just frozen screen with "space invader" looking artifacts for a second, then infinite backscreen. Most of the crashes I encounter, besides blue screens, don't fully shut down the computer, as I can hear sound from whatever app I was using before, if that counts for anything.

The only artifacts I ever get look like "space invaders", and on rare occasions after a hard BSOD it will show tiny purple lines on the boot screen and the pc will have trouble booting until I turn it off and on again. I ran a test on it using MATS, and the vram came back fine and passed it (I ran this once). I do have micron ram unfortunately.

Not my picture, but this is generally what I see a split second before I black screen, except that they are more limited to the middle of the screen (horizontally vertically), and vary in color depending on the placement of them. I haven't gotten a picture myself because it is pretty quick, but the artifacts resemble the shape I see.

My picture, this is upon rare occasions after a BSOD. It usually ends up freezing at this page.

GPU-Z screenshot


I did however open up the gpu and discovered that there was barely any thermal paste/it was just worn out and hard, so I went ahead and replaced that, and my temps are a bit better, and it made my card last a bit longer before any crashing. It might be worth to note that the gpu hotspot is generally around 14 C above the core, where previously it was 20 C + above. The thermal pads looked old as well so I went ahead and ordered some new ones, and I should be able to throw them in soon to see if that fixes my issues.

While running this card, and putting it under load, my psu likes to whine pretty hard. With my EVGA 1080ti sc, I don't notice a whine as much. I do notice that before the 2080ti crashes the psu likes to kick in a little harder and I start to hear it more.

If this card hasn't been touched at all for a certain amount of time, I can casually run furmark, multiple gpu accelerated programs (chrome, spotify) and a game at the same time without a problem until the first crash happens. ( I was trying to really push it to see if a heavy load was really the issue )

As for what I think it mainly is, is overheating. I repasted my gpu with the only thermal paste I had on me to see if it did anything (Arctic MX-4) and it did make the card last a bit longer. Now, since I am still crashing, I think my vram is getting too hot, but I have no clue if it is or not as my card does not have memory temp sensors. I'll post an update when my thermal pads come in and what they end up doing.

I have not ruled out that it could be my psu, though. Hearing it work pretty hard before it starts to crash concerns me that it might be my psu but I have done no tests to prove so. I do find that putting fans on full blast after a crash will cause it to crash, fans would speed up again, then crash. Rinse and repeat till it hard crashes.

I haven't experienced any of these issues running my EVGA 1080ti sc.

So those are just my experiences/thoughts about the issues I am dealing with while running this 2080 Ti FE. I do know that a lot of the time it is just the micron ram that is bad or is going bad, but I would like to explore my options before I bin the card, as I can not return it. I will appreciate if any of you guys with some experience could try and help me with this issue.

 

[thestryker]

It's sounding like a power/temperature issue a bit more than necessarily a bad card issue. Were the thermal pads on VRM/memory in good condition when you took the cooler apart? Your PSU should be sufficient for the card, but I'm wondering if you ran two separate cables for GPU power or are using a single daisy chained one. I'm not sure what all that PSU comes with cables wise, but I think it's one single and one double.

 

[1vk3]

It's sounding like a power/temperature issue a bit more than necessarily a bad card issue. Were the thermal pads on VRM/memory in good condition when you took the cooler apart? Your PSU should be sufficient for the card, but I'm wondering if you ran two separate cables for GPU power or are using a single daisy chained one. I'm not sure what all that PSU comes with cables wise, but I think it's one single and one double.

Thank you for a quick reply. The cables aren't daisy chained and are separately plugged into the psu.

As for the thermal pads, they were ripped upon taking the card apart and kind of mushy, pretty stretchy. I just put the parts back where they were ripped, which is probably stupid, but it was the best thing due to having no excess thermal pads.

 

[thestryker]

Thank you for a quick reply. The cables aren't daisy chained and are separately plugged into the psu.

As for the thermal pads, they were ripped upon taking the card apart and kind of mushy, pretty stretchy. I just put the parts back where they were ripped, which is probably stupid, but it was the best thing due to having no excess thermal pads.

Definitely the right way to hook up the GPU when it's an option so nothing wrong there.

Honestly as long as enough of them was left it might not be quite as bad as it might seem. You should probably look into replacing them though as a long term solution. You'd have to look up what thickness they are just to make sure you get the right kind, or potentially even get a paste meant to take the place of pads.

You could try doing a bit of power limiting with MSI Afterburner. I think the FE should be 250W (which is the same as the 1080, but excursions might be higher) so maybe try 80% power to cut it down to ~200W and see if that makes any difference. It would also be worth looking at power consumption in Afterburner just to make sure what it is actually pulling. If you don't have crashes and the PSU isn't making as much noise you may need to try a different PSU. It is however possible that there's a power delivery issue with the card itself.

 

[IDProG]

Thank you for a quick reply. The cables aren't daisy chained and are separately plugged into the psu.

As for the thermal pads, they were ripped upon taking the card apart and kind of mushy, pretty stretchy. I just put the parts back where they were ripped, which is probably stupid, but it was the best thing due to having no excess thermal pads.

Unlike the others, I am not going to pull my punches.

- You got tricked
- You bought a post-mining used card,
- The WORST types of cards to be bought used, mind you (GDDR6X)
- Miners overclock the GPU's VRAM like crazy and leave it at, like, 110°C for years. If you don't know, that's the reason why the artifacts show up. It's because of VRAM producing errors, not being able to store graphics properly.
- You just bought an e-waste and wasted $300 for it. Well done.
- The miner is currently laughing at you so hard to the bank

Now that you have learned the lesson the hard way, hopefully you know better next time.

 

[Tac 25]

have to agree with @IDProG .

looking at the symptoms described at the first post. This sadly, looks like a VRAM issue.

 

[SkyNetRising]

Dead card.
VRAM chips need to be resoldered or replaced.

 

[1vk3]

Definitely the right way to hook up the GPU when it's an option so nothing wrong there.

Honestly as long as enough of them was left it might not be quite as bad as it might seem. You should probably look into replacing them though as a long term solution. You'd have to look up what thickness they are just to make sure you get the right kind, or potentially even get a paste meant to take the place of pads.

You could try doing a bit of power limiting with MSI Afterburner. I think the FE should be 250W (which is the same as the 1080, but excursions might be higher) so maybe try 80% power to cut it down to ~200W and see if that makes any difference. It would also be worth looking at power consumption in Afterburner just to make sure what it is actually pulling. If you don't have crashes and the PSU isn't making as much noise you may need to try a different PSU. It is however possible that there's a power delivery issue with the card itself.

After a bit of thorough testing (aka running a minecraft instance with shaders, furmark, and streaming services at the same time) I fixed it! I don't know how the thermal pads really helped but they sure did. No crashes or tiny stutters, better temps and framerates.

I did find that even though the old thermal pads were worn out, I didn't do an amazing job putting them back either after they fell off the first time around.

Thank you for your insight!

Unlike the others, I am not going to pull my punches.

- You got tricked
- You bought a post-mining used card,
- The WORST types of cards to be bought used, mind you (GDDR6X)
- Miners overclock the GPU's VRAM like crazy and leave it at, like, 110°C for years. If you don't know, that's the reason why the artifacts show up. It's because of VRAM producing errors, not being able to store graphics properly.
- You just bought an e-waste and wasted $300 for it. Well done.
- The miner is currently laughing at you so hard to the bank

Now that you have learned the lesson the hard way, hopefully you know better next time.

Did you even look at what I was dealing with? 😛

 

[thestryker]

After a bit of thorough testing (aka running a minecraft instance with shaders, furmark, and streaming services at the same time) I fixed it! I don't know how the thermal pads really helped but they sure did. No crashes or tiny stutters, better temps and framerates.

I did find that even though the old thermal pads were worn out, I didn't do an amazing job putting them back either after they fell off the first time around.

Thank you for your insight!

Glad you were able to square it away as it didn't seem like the card was bad per se just something was hampering it.

Did you even look at what I was dealing with? 😛

It has become somewhat of common knowledge that artifacting means memory issues. The problem with that is memory artifacting is very specific and it's the textures themselves where the artifacting lies and a lot of people just aren't aware of that. I'd seen it first hand for the first time about 20 years ago and really if you haven't seen it personally I can see where the confusion comes in.

 

[IDProG]

Did you even look at what I was dealing with? 😛

Hmm, you probably thought I meant bad. You probably thought that I didn't want you to have a happy ending with the card.

No. If your card is currently working as usual without error right now, props to you. I have nothing against you. Genuinely.

Just know that if it doesn't sometime later, hopefully you understand the reality of the situation.

It has become somewhat of common knowledge that artifacting means memory issues. The problem with that is memory artifacting is very specific and it's the textures themselves where the artifacting lies and a lot of people just aren't aware of that. I'd seen it first hand for the first time about 20 years ago and really if you haven't seen it personally I can see where the confusion comes in.

No. I could say that because I knew it first-hand. It's not that I bought a post-mining used card (that would've made me a hypocrite if I did).

I overcloked my GPU's VRAM too high and it produced similar looking artifacts. That's how I know it. It didn't happen with me overclocking the core clock too high, only VRAM.

 

[thestryker]

No. I could say that because I knew it first-hand. It's not that I bought a post-mining used card (that would've made me a hypocrite if I did).

I overcloked my GPU's VRAM too high and it produced similar looking artifacts. That's how I know it. It didn't happen with me overclocking the core clock too high, only VRAM.

Except that the original poster put their own picture up which clearly shows it isn't VRAM artifacting. You made a false assumption because you conflated the two not knowing what you were looking at.

 

[IDProG]

clearly shows it isn't VRAM artifacting.

Are we looking at the same pictures? It was CLEARLY VRAM artifacting. That is exactly how VRAM artifacting looks like.