Question BSOD with RTX 2080 Ti (need help)

[crispy.cc]

Hey everyone, I'm having some trouble with my RTX 2080 Ti which has (seemingly) died for some unknown reason 🙁

For two days I've been trying to troubleshoot sudden BSOD crashes on my PC.
This has never happened on the system before, with exceptions for some corrupt drivers that have been fixed occasionally.

What I've experienced
1. PC crashes under high GPU load, anything from 10 seconds to 10 minutes.
2. Application crashes under high GPU load (games, benchmarks etc.) but no BSOD.
3. PC does not crash at all during regular use (low GPU load)

What I've tested
1. Different PCI-e slot
2. Different Displayport output
3. Different Displayport port on GPU
4. Different Displayport cable
5. Complete driver wipe with DDU
6. Nvidia debug mode
7. Entirely different PC with same GPU

8. Use my old GTX 980 (this works)

This is what the crashes look like

Dxdiag, sysfilecollection, perflog, gpu-z log, disassembled card images here:

2080Ti Issues – Google Drive

drive.google.com

System specs
Intel Core i7-4790K
EVGA RTX 2080 Ti SC
ASUS Z97 Maximus VII Ranger
Corsair Dominator Platinum DDR3 16GB (4x4)
Corsair HX850i 80+ Platinum
Corsair LX256GB SSD
Seagate Barracuda 2TB
(Corsair H105i CPU Cooler)

Original OS: Windows 7 64-bit
Current OS: Windows 11 64-bit
Build date: 2015, upgraded GPU March 2020
The latest Nvidia driver 537.42 was released 2023-09-21, which is more than 10 days before the crash.

The GPU is 7 months past its warranty period and unfortunately EVGA is not willing to create an RMA case for me.

Any help is appreciated as it would cost me a small fortune to replace the card.

 

[Phaaze88]

This is what the crashes look like

Gpu memory is overheating, or has kicked the bucket. 2080Ti was prone to memory failures.
I don't suppose hwinfo(sensors only) or Gpu-Z(sensors tab) tells you the operating temperature of the Vram?

 

[crispy.cc]

Gpu memory is overheating, or has kicked the bucket. 2080Ti was prone to memory failures.
I don't suppose hwinfo(sensors only) or Gpu-Z(sensors tab) tells you the operating temperature of the Vram?

Unfortunately no, only tells me the GPU temperature and hot spot, i've never seen the hot spot go above 85 degrees.

 

[Phaaze88]

Unfortunately no, only tells me the GPU temperature and hot spot, i've never seen the hot spot go above 85 degrees.

Ah, so much for that suggestion. About the only way to check memory temperature from there is to use thermocouples and get a reading off of that... or you could try removing the backplate of the gpu, blasting the back of the card's memory area with a fan, and seeing if it prevents the artifacts.

Hot spot is the warmest sensor reading from the gpu die; there are multiple sensors on die. It's not for memory.

 

[crispy.cc]

Ah, so much for that suggestion. About the only way to check memory temperature from there is to use thermocouples and get a reading off of that... or you could try removing the backplate of the gpu, blasting the back of the card's memory area with a fan, and seeing if it prevents the artifacts.

Hot spot is the warmest sensor reading from the gpu die; there are multiple sensors on die. It's not for memory.

Seems the VRAM is common issue with the 2000-series cards. Shame mine broke after it's warranty period.

I have ordered a RTX 4080 for replacement. I'll be living off noodles for a couple of months to deal with the financial burden of a 1550eur card though...