Video card problem?

[iamjn]

Hi all. Have a problem that I'm scratching my head over. Looking for another opinion.

I was due for a Windows refresh anyways, since that build was pretty old and running like garbage due to some unrelated issues, so thought I'd throw in a spare SSD that I had on hand. Installed the SSD with clean install of Windows 7, then upgraded to 10. The main specs on this older PC are:

i5-3470
12GB DDR3
AMD 6870 Eyefinity 2GB
440GB SSD
3TB HDD (storage)
1TB HDD (storage)

The main use is desktop apps, network management apps for work (when I'm feeling too lazy to pull out my laptop when I'm on call), maybe an occasional game, but nothing heavy or newer on this machine, as that's what the other one is for.

Anyways, after the upgrade to 10 I noticed that anytime I did anything that put any load on the GPU, I'd get the 'Driver has stopped responding and recovered' error in Windows. If I try to run something that really pushes it, I'll get some momentary tearing, then a BSOD As it's a legacy GPU, I was running the last driver from AMD. They released a beta Crimson driver for it, which installed without issue. Still got the error and crash. I tried reseating the card, verified the power (thought maybe I knocked it out moving drives around), messed around with the TdrDelay some, underclocked it. Same issue. Thought perhaps it was a problem with Win 10 so did a clean Win 7 install again, and have the same problem still. Thought that maybe I was drawing too much power with the additional drive, so unplugged the spinning disks and just ran on SSD-same problem. I'm leaning towards the card just died and needs to be replaced, but given that I can run all day long using desktop apps kinda makes me think that it could still be something else. Any thoughts? I'm not opposed to replacing the card, but for what I use this PC for, I may just dump the whole system instead.

 

[LTC8K6]

Gpu-z to see what the card temps and voltages look like?

 

[iamjn]

GPU Core Clock [MHz] 300
GPU Memory Clock [MHz] 1050
GPU Temperature [°C] 56
Fan Speed (%) [%] 44
Fan Speed (RPM) [RPM] -
GPU Load [%] 9
GPU Temp. #1 [°C] 56
GPU Temp. #2 [°C] 91.5
GPU Temp. #3 [°C] 57.5
Memory Usage (Dedicated) [MB] 131
Memory Usage (Dynamic) [MB] 33
12V [V] 12.06
VDDC [V] 0.945
VDDC Current [A] 4
VRM Temperature 1 [°C] 24
VRM Temperature 2 [°C] 25

 

[LTC8K6]

91.5C sounds kind of hot...
Not sure what that temp is related to, though?

 

[iamjn]

Agree. Unfortunately I can't find much on that sensor. It definitely doesn't fee that hot.

 

[LTC8K6]

Try leaving the case open with a fan blowing into it to see if the problem goes away?

 

[Ketchup]

The question is, is the card hot because it's dying, or hot because it isn't getting good airflow/contact with the heatsink.

The fan idea isn't bad, but only a temporary solution for the heat. Get the dust off the video card, remove the heatsink, and re-apply thermal paste on the GPU. If it is still overheating, it's probably a bad card.

What board do you have? It probably has video out, and since your CPU does has video built in, you can run the computer off that (video card removed) to make sure the video card is what was causing your issues before replacing it.

As for your testing so far (Windows 10, Windows 7, different drivers), you have pretty much covered everything on the software side for such a situation.

 

[Burpo]

Get the dust off the video card, remove the heatsink, and re-apply thermal paste on the GPU. If it is still overheating, it's probably a bad card.

+1