Question Games crashing RTX 2080ti - GPU Z 'hot spot' Temperature high

desecravity

Distinguished
Nov 6, 2012
65
0
18,660
6
Hey all,

I've got a Seahawk 2080ti on an EK waterblock that I seem to be having issues with. The PC was built in 2019 and I have never had any issues until recently.

Every time I play demanding games I inevitably end up crashing. Certain games like the the new MW2 will crash very quickly, and others will play for longer periods and then crash - I'm assuming that the more demanding games are quicker to crash in this instance. I don't crash in any other applications.
The crashes are always the same - Monitor loses signal, sometimes the game sound loops, all PC lights and fans running like normal. I always have to hard reset to get back into Windows. Benchmarks like Timespy and Heaven usually run without crashing (I'm guessing they're just not as demanding as MW2 at 3440x1440)

I have tried all the standard trouble shooting stuff - uninstall driver with DDU, then reinstall. Malware/virus scans. Set GPU to stock clocks. Underclocked GPU. I even did a fresh installation of Windows.

I initially thought my GPU was dying, but the more I think about it, the more I think I might be dealing with a temp issue. I checked on GPU - Z that my GPU temp is running at a max of 70C, but what caught my eye was that my 'hotspot' temp was at 107C. Could it be that I need to remove the water block and reapply thermal paste?
It might sound like a silly question, but I have seen on another post that the 'hotspot' feature is not reliable on the 2000 series cards and that the temp will be a lot closer to what I am actually getting through the overall monitor i.e. 70C?

Any advice appreciated, as I am pretty much on my last leg regarding options. Only thing I can think of doing is reapplying thermal paste or replace GPU - and what a bad time to be doing this imo.
 

Phaaze88

Titan
Ambassador
If the hot spot is getting that high, but the average isn't up there with it, that usually points to the block not being installed correctly, or warped mounting pressure - no fault of the paste.
Gpu temperature reading is an average. Hot spot is the hottest sensor on the die.
 

desecravity

Distinguished
Nov 6, 2012
65
0
18,660
6
If the hot spot is getting that high, but the average isn't up there with it, that usually points to the block not being installed correctly, or warped mounting pressure - no fault of the paste.
Gpu temperature reading is an average. Hot spot is the hottest sensor on the die.
Thanks for the input. Only strange thing is that the build has been untouched since 2019 and working as it should - so to me that kind of rules out the block being mounted incorrectly. The GPU is also mounted vertically to a bracket with an adaptor so there is no pressure on the PCI lanes - that would also rule out the warped mounting pressure imo.
Do you happen to know if 'hotspot' temps is accurate with 2000 series?
 

ASK THE COMMUNITY