Question Used RTX 3080 GPU having high hot spot temperature and crashing

[Amir3022]

I recently got a used Palit Gamerock RTX 3080 10 gb from a local store. Run the usual tests and everything looked fine temps and performance no artifacts as well in any of the tests I ran. Anyway once I installed the th GPU in my system the GPU would crash after 1.5 or 2 hours of gaming or heavy load, with fans occassionally ramping up all the way to 3500+ RPM, and also when it crashed. I used to watch temperature using MSI Afterburner, but suspected maybe more is to it. Downloaded HWInfo and saw that GPU hotspot surpasses 105°c when the fans ramp, which leads to core clock decrease to base clock and sometimes crashes the GPU. Tried decreasing the GPU power limit to 50% which stopped the crashing even after 5+ hours of heavy load, fans stopped ramping and hot spot temperature maxed at 87°c.
So what could be causing this? Would a repaste be enough to solve this problem? Or could it be a faulty GPU and I should be trying to get ky money back? All help is appreciated
System Specs:
MB: Gigabyte b650 Gaming AX
CPU: AMD Ryzen 5 7600x
Mem: 1x16gb DDR5 5200mt/s
GPU: RTX 3080 10gb Palit Gamerock
PS: Ximatec 750w 80 plus silver

 

[Phaaze88]

You really only have 1 dimm of ram? You gotta fix that. That cuts cpu performance by quite a bit - can't give you a concrete %'age, because it varies. I just know that it's an ouchie if playing anything recent.
Return(if it's still within the return window), or sell that kit and get a proper 2 dimm kit. You could just buy another dimm, but there's a bit of lottery involved in that route.

Or could it be a faulty GPU and I should be trying to get ky money back?

If you can, absolutely do that first. This shouldn't be your problem to fix, especially not with a 2nd hand gpu.

So what could be causing this?

Previous owner:
-did a repaste, but didn't do it correctly, OR it could have come like that from the factory; crap happens sometimes. In which case, said owner should've RMA'ed it themselves, so I'm leaning towards the former reason.

-changed out the thermal pads with the wrong ones, which is very easy to screw up. Pad replacement shouldn't be done unless absolutely necessary. They're expensive, and it's possible to go through a few sets before you get the right ones.

Would a repaste be enough to solve this problem?

Might not be. See the part above about thermal pads.

 

[Amir3022]

You really only have 1 dimm of ram? You gotta fix that. That cuts cpu performance by quite a bit - can't give you a concrete %'age, because it varies. I just know that it's an ouchie if playing anything recent.
Return(if it's still within the return window), or sell that kit and get a proper 2 dimm kit. You could just buy another dimm, but there's a bit of lottery involved in that route.

I am planning on getting another dimm, I just built this PC recently and i was short on money, so I went with a single DIMM for now, but certainly planning on getting another exact model in the near future.

As per the GPU, I contacted the seller, and he told me the GPU was never opened, and that shows since the seal is still present on the screws of the GPU. He agreed to pay for thermal paste change, and if that doesn't fix the issue, I will get my money back.

I will update this thread once I do this and re-test the GPU and check if the problem is fixed.

 

[Phaaze88]

I am planning on getting another dimm, I just built this PC recently and i was short on money, so I went with a single DIMM for now, but certainly planning on getting another exact model in the near future.

As per the GPU, I contacted the seller, and he told me the GPU was never opened, and that shows since the seal is still present on the screws of the GPU. He agreed to pay for thermal paste change, and if that doesn't fix the issue, I will get my money back.

I will update this thread once I do this and re-test the GPU and check if the problem is fixed.

Mmmk. Fingers crossed then.

 

[Amir3022]

Mmmk. Fingers crossed then.

So I just repasted the GPU and changed the thermal pads, and re-tested the entire testing suite, and no it works pretty well! Average temperature reading doesn't exceed 65c and the hotspot reaches a maximum of 83c under heavy load and full powered GPU. The fans don't ramp up as well as they used to do. I ran a 5+ hours test and it didn't crash at all and temperatures are reasonable. So I guess it was a thermal problem and mostly solved. I will update this thread again a week from now after a few days of load and report how is everything working.