Question Is my 2080ti FE actually dead?

[yourcoastie]

System:
Prime Z390-A Motherboard
i9-9900k
16gb RAM
2080ti Founders Edition
Corsair HX1000i Power Supply.

My GPU's been causing purple streaks and artifacts when it crashes( not sure if that's relevant). It also ONLY ever happens under gaming load. But when it started happening, I started monitoring MSI Afterburner. It reached like 65C and crashed. Tried lowering both Core Clock and Memory Clocks down, still same thing. Making my fans spin at max percentages and nothing.
I have a spare GTX780 and borrowed a 4070ti to make sure it's my card causing the issue.

So I ended up thinking its a thermal issue and specifically bought thermal paste and thermal pads to redo it all.
https://www.amazon.com/dp/B011F7W3LU?psc=1&ref=ppx_yo2ov_dt_b_product_details - Thermal Grizzly Kryonaut
https://www.amazon.com/dp/B096ZNHY8F?psc=1&ref=ppx_yo2ov_dt_b_product_details - OwlTree branded thermal pads

After I was done it ran good as new. Cooler too....for 2-3 days. Now I'm back in the exact same spot where it crashes under load anywhere at 55C+

Do you think it's anything specific or is the lifespan actually over for my card?

Edit: Seemed to have fixed it by increasing Thermal pad thickness specifically on VRM from 0.5mm -> 1mm.
Thank you for helping out. This forum is so nice.

 

[Phaaze88]

There are old articles mentioning 2080Tis being prone to vram failures. Could be that - but since you said this:

After I was done it ran good as new. Cooler too....for 2-3 days.

Now, I'm not so sure. Maybe something started overheating again. What's the gpu hot spot like? [FYI, Afterburner does not show that one. Gotta use hwinfo or Gpu-Z.]

What's the make and model of that power supply, and how long has it been in use?

 

[yourcoastie]

There are old articles mentioning 2080Tis being prone to vram failures. Could be that - but since you said this:

Now, I'm not so sure. Maybe something started overheating again. What's the gpu hot spot like? [FYI, Afterburner does not show that one. Gotta use hwinfo or Gpu-Z.]

What's the make and model of that power supply, and how long has it been in use?

This is what hwinfo reads. Note, this is after I heated it up just now to force a crash to get it hot.

https://imgur.com/a/zqga4q0

View: https://imgur.com/a/zqga4q0


Also my psu is a Corsair HX1000i. I've had it for maybe 2-3 years. Corsair sent a replacement after the first one died on me.
Edit: I have a CX750M that I troubleshooted with before redoing the thermal pads/paste. It would crash the same way.

 

[Phaaze88]

Nothing unusual there, and the psu is good.

Perhaps try a repaste and see if it does the same thing after a couple days? Maybe the 'good for 2~3 days' was a fluke?

 

[stonecarver]

View: https://www.youtube.com/watch?v=t5memuI5WD4&t=354s


Maybe you can limp the card along with info in this video. Watch the whole thing but if you want what he did is at 4:25 into video.

 

[thestryker]

Did you measure the pads or find the measurements to make sure you had the right thickness in the right areas? It's possible poor contact on the pad replacement is why it worked okay for a bit before causing problems again.

 

[yourcoastie]

Did you measure the pads or find the measurements to make sure you had the right thickness in the right areas? It's possible poor contact on the pad replacement is why it worked okay for a bit before causing problems again.

I actually followed this guide for that

https://hardforum.com/threads/replacement-thermal-pads-for-2080ti.1981252/

I'm taking it apart to maybe put 1 layer thicker on the VRM

 

[thestryker]

I actually followed this guide for that

https://hardforum.com/threads/replacement-thermal-pads-for-2080ti.1981252/

I'm taking it apart to maybe put 1 layer thicker on the VRM

Are the pads you got really soft? When I redid my current card I used thermal putty, but the pads on it were much softer than the pads I had left over from another build.

Definitely check imprints on the pads when you take it apart again just to see if something is perhaps too much or not enough. I really wish the makers just provided measurements/part numbers as it seems like the pads always go first. The putty I used isn't sold anymore and I haven't seen a good replacement for it unfortunately so I don't have any suggestions for you there as an alternative.

 

[yourcoastie]

Are the pads you got really soft? When I redid my current card I used thermal putty, but the pads on it were much softer than the pads I had left over from another build.

Definitely check imprints on the pads when you take it apart again just to see if something is perhaps too much or not enough. I really wish the makers just provided measurements/part numbers as it seems like the pads always go first. The putty I used isn't sold anymore and I haven't seen a good replacement for it unfortunately so I don't have any suggestions for you there as an alternative.

They are soft. Just barely firmer than the original pads. I wouldn't call them putty and moldable though

 

[yourcoastie]

Update: Im not sure my VRM's are making contact because there's almost no indentation?

https://imgur.com/a/oh6Jaxu

View: https://imgur.com/a/oh6Jaxu

 

[stonecarver]

Well a little to much thermal compound for starters.

I once fixed a 8800 GT that always ran hot and took it apart and was just frosted like a sloppy cupcake on GPU core. Cleaned and with new compound the card ran like it should after.

I see what you mean it does seem the pads on memory could be to thin or the pads everywhere else are to thick keeping pads on memory to far away.

 

[thestryker]

Update: Im not sure my VRM's are making contact because there's almost no indentation?

With the way it looks on yours it's almost as though the 1mm and 0.5mm from the HardOCP forum post should be reversed since all of the ones listed as 0.5mm on that post are the ones with the least imprint left.

 

[yourcoastie]

With the way it looks on yours it's almost as though the 1mm and 0.5mm from the HardOCP forum post should be reversed since all of the ones listed as 0.5mm on that post are the ones with the least imprint left.

You have a sharp eye. I upped the VRM pads from 0.5mm to 1mm and it all seems to be back to normal. Well except my gpu temp is a little higher. But I think that's from opening it up twice after repasting. Also too much paste like someone mentioned above. Thank you for the help!

 

[yourcoastie]

Looks like just over a month of completely fine running, I'm back to the same issue.

Is there a more efficient way to cool VRM? Apparently the pads I used had a thermal conductivity of 6.0W/mK so I bought some with 13 W/mK, coping that it'll be what can fix this again.

Edit: after cutting the old pads instead of 1 strip for 4 VRM's, I made it 1 square for each and it's back to normal? I don't know what's going on anymore.

Edit2: Back to crashing again. I think the pads are not pressing up to the VRM as tightly as they need. Because when I take it back apart, they're not stuck to the VRM, they're stuck to the backplate.

Edit3: When I was redoing the thermal pads and placing them directly to the pcb, I noticed I completely missed one long strip of padding right next to the VRM. I think this is actually why it was running alright but after some time eventually that strip got real hot.

tldr; I'm stupid and didn't notice I was missing a strip of thermal pad.