My 2080 ti is showing it's age... maybe?

[MuchachoSal]

Howdy all, I've had an ASUS 2080 ti for about two years (this one: https://www.amazon.com/gp/product/B07GJYS2WM/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1) and for more or less, it's worked out pretty well.   I had some fan trouble back in July that caused some massive throttling but I was able to use some machine oil to grease the bearings - everything went well after that.

 

Now, I have what sort of seems to be a new problem that's only crept up within the last three weeks, although I wouldn't be surprised if it's related to the fans.   In short - the card now crashes when it's "pushed to the limit".   Games like Overwatch, Assassin's Creed Odyssey, Horizon Zero Dawn, Control, etc. - all games that can push the limit - crash around the time that the card reaches 69-70c.  In the past, I've been able to run them at 1440p (~60 fps, whatever) with max settings on each without much problem, but anytime I do that now, the games crash pretty quickly after starting up - the renderer crashing is the error that I get almost 100% of the time now.   

In order to prevent crashing, I have to force a 60 fps limit and also dial down the internal resolution to around 1080p.  I also have to use MSI to crank up the fans to 100%.  Even then, I still get some crashes.  I've reloaded windows, I've gone to previous Nvidia drivers, etc.   Same problems.   It's really perplexing.     

 

The 1.png screenshot shows GPU-Z while running Horizon Zero Dawn with 1152p resolution.  The black box at the end is when the game crashes.  I noticed that the sole Perf Cap reason switches to Pwr (Green) before the crash, which is something that I saw similarly when I had my fan problem back in July.  Also - and this is hard to tell from the screenshot, so I've attached the log -  Fan 2's speed, just right before the crash, drops about 2.5% in relation to Fan 1's speed.   When the speed (slightly) drops, the temp and gpu load both go up to the point where the crash takes place (around 16:59:11).

 

So, something in my card is basically saying "nope, no can do".    My fans *seem* to be fine.   I'd be open to getting a new cooling system for my card but I honestly wouldn't know how to pull that off without one of Linus' hacks, and even then I'm not confident it would fix the real problem. Any ideas here?  Need any more info?

 

Thanks in advance!

 

GPU-Z Sensor Log.txt

[Moonzy]

Warranty...?

Tried DDU and reinstalling the drivers?

[Grabhanem]

Have you tried something like Furmark to isolate the problem to GPU load?

[xg32]

32 minutes ago, Grabhanem said:

Have you tried something like Furmark to isolate the problem to GPU load?

i wouldn't recommend this, it might push his gpu off a cliff, if he cant game normally after a ddu or windows install that's enough evidence imho.

 

@MuchachoSal if it's related more to temps and not load, u can try taping or clipping on a big 3000rpm 120mm fan (on top of the faulty fan) and pull air from it to prevent it from ever getting to 70C (janky but cheap solution, squeezed an extra year out of my 970 with very low temps), if that doesn't work, time to hit up warranty 

[GamerDude]

Two years, eh? Have taken the card out to look at the heatsink? I mean after two years, the heatsink may have accumulated dust or dust bunnies that may impede proper cooling. 

[MuchachoSal]

So, a few more things:

 

1. I know the CPU isn't the problem.  Last week, I removed the 19% overclock I had on my 10600k and I ended up having the same problem.

2. I just ran Furmark. (before seeing xg32's post, Oh well )  The first time I ran a 1440p benchmark it crashed at 80% finished.  The second time it didn't crash at all   I've never actually ran Furmark before.  Reminds me of the the Eye of Sauron, but a cuddly Eye of Sauron.  Interestingly enough the GPU temps didn't get that high before the crash took place. 

3. I'll look at the heatsink - it might need some TLC.  Like I mentioned earlier, I did fix one of the fans so I have done at least some operations before. 

4. And sure, I'll see what I can do about pulling more air via another fan.  And then warranty as a last resort.  Heck, I'd even buy a 3080 if there was one available at retail price right now.

 

Thanks for the tips all!

[GamerDude]

2 hours ago, MuchachoSal said:

So, a few more things:

 

Heck, I'd even buy a 3080 if there was one available at retail price right now.

Good luck with that!!! 

[Froody129]

Sounds to me like memory or some power component overheating.

 

As a last case I’d try the card in a different PC or a different power supply. Might even be your RAM and not the graphics card at all. 
 

I’d just send it in for warranty repair if it still has one. Clearly a manufacturing defect since the card is so new still. 

[Ebony Falcon]

Ddu reinstall

msi afterburner custom fan curve 

max power and voltage slide and dial back ur normal oc by 20mhz unless ur already at stock 

give that a go 

 

keep eye on temps ect

[HumdrumPenguin]

Remove one memory stick, do the test and see what happens. Then use the stick you've just removed by itself on another test run. Report. This will troubleshoot the RAM in case that's the problem. Also, what PSU are you using, and for how long have you used it?

[martward]

Normally I'd say RMA the card, but I wouldn't be surprised if throwing machine oil at it voided the warranty. I'd start DDU'ing the drivers, then I'd check for problems with the RAM and the PSU, since I don't see crazy temperatures in that screenshot. If nothing works you can try to throw an aftermarket cooler on there. 

For example:

https://www.arctic.de/en/Accelero-Xtreme-IV/DCACO-V800001-GBA01