• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

GPU Hot spot temps too high on 6800XT?

Justnotsleeping

Justnotsleeping

Justnotsleeping

Associate
Joined
27 Sep 2020
Posts
147
Hi,

Im looking for some advice on my GPU. I have never really had any issues with it, but I have been playing through the campaign of CoD:Modern Warfare II during last few days. The fans have been quite loud but this is not unusual when playing fairly demanding games, and not sure that the Gigabyte gaming OC model has the best of coolers on it. I play on a 3440 x 1440p ultrawide. The CPU is cooled with an NZXT AIO and temps are great, and there is also plenty of airflow in the case.

What was unusual is that I have been experiencing quite a few crashes in the game where the screen goes black, the fans drop down and i have to hard reset the PC.

Id noticed that the GPU utilisation had been at 99% which I guess might not be that unusual but because of the crashes I decided to run HW Monitor so I can check the GPU temps.

GPU Temperature

Min:- 39C
Max:- 79C
Average:- 70C

Memory Junction

Min:- 42C
Max:- 76C
Average:- 68C

GPU Hot Spot Temp

Min:- 42C
Max:- 109C
Average:- 95C

The GPU Temp and Memory Junction don't seem too concerning, but the the Hot Spot Temp hitting close to 110C seems very high. I know this will be the hottest part of the card, but even so this seems quite high, and I wondered whether this was contributing to the crashing.

Has anyone noticed extremely high temps on this card, either generally or whether this might be an issue with this game only. I've never had any game crashes running games like Cyberpunk, Hogwarts, etc.


You think its worth a repaste of the GPU... I already have a tube of Thermal Grizzly Kryonaut so I dont have to buy any, but will this have little impact?

Thanks
 
GPU utilization @ 99% is good - it means your CPU isn't bottlenecking your GPU.

I don't have a 6800XT myself but according to Gamers Nexus' review, they do seem to run pretty hot (although that 109° does seem extreme).

www.youtube.com

AMD Radeon RX 6800 XT GPU Review: Gaming, Thermals, Noise, & Smart Access Memory Benchmarks

This is our review of the AMD RX 6800 XT video card, looking at RDNA2 and the reference cooler vs. the RTX 3080, 3090, 3070, RX 5700 XT, & more. Includes Sma...
www.youtube.com www.youtube.com
 
Justnotsleeping said:
and not sure that the Gigabyte gaming OC model has the best of coolers on it.
Click to expand...
Had the 6800XT GamingOC for a short time, found it to be v good. Hotspot is a little high though, think mine had like a 12/13c delta between edge and hotspot. I'd just tweak the settings/undervolt so it never reaches the 110c point (where the card starts to throttle).
 
Justnotsleeping said:
Its just at stock. I see a lot about overvolting GPU/CPU but its not something I've tried. I'll maybe lookup a guide..
Click to expand...
Not overvolting, undervolting. Slightly lowering the volts can actually be seen to show performance increases as lower volts = lower temps = longer/higher boost performance (to put it in its most basic form)
Well worth having a look at some YouTube guides and may well fix your problem.
 
Yeh did this with my 6700 on the weekend (re undervolt / overclock).

Found an easy to follow guide with a Google search. And in Adrenalin, it was very easy to do, especially so as the GUI is very user-friendly.

3DMark score went up with a small, but nice improvement and the card is now rated as "Great" rather than "Good" (guess I was on the cusp anyway but it made for good reading!).
 
Thekwango said:
Not overvolting, undervolting. Slightly lowering the volts can actually be seen to show performance increases as lower volts = lower temps = longer/higher boost performance (to put it in its most basic form)
Well worth having a look at some YouTube guides and may well fix your problem.
Click to expand...
Had a couple of crashes now in Diablo 4 to which I wouldn't have thought was particularly GPU intensive. Hotspot temp wasnt quite 110 C though, at this time max was about 105C. I'll find some Youtube guides on undervolting in the Adrenalin software to see if it sorts the potential problem.

Trying to think if any other configurations have changed on my PC but nothing significant springs to mind....only recent non gaming software I installed was Rainmeter to customise my desktop. I had used it previously though and dont recall any crashing.

Will go down the undervolting route, and might just repaste the GPU anyway, figure it cant do any harm.
 
Justnotsleeping said:
Had a couple of crashes now in Diablo 4 to which I wouldn't have thought was particularly GPU intensive. Hotspot temp wasnt quite 110 C though, at this time max was about 105C. I'll find some Youtube guides on undervolting in the Adrenalin software to see if it sorts the potential problem.

Trying to think if any other configurations have changed on my PC but nothing significant springs to mind....only recent non gaming software I installed was Rainmeter to customise my desktop. I had used it previously though and dont recall any crashing.

Will go down the undervolting route, and might just repaste the GPU anyway, figure it cant do any harm.
Click to expand...
On my 6800 I just did -50mv first and that was solid. Haven't done the rest yet but will be taking 25 mv off at a time until I feel it's enough or it crashes and then I'll just add a little more back and test. Plenty of good tests out there if you google.

Repasting normally always lowers temps if done right but its just the warranty stuff. Most are ok with it though.
 
Last edited:
By all means try some under volting. However the difference between the max hot spot and the max normal temp is quite large. That normally means the thermal paste has degraded. I think it’s time to re paste the card. If it’s still under warranty you might be able to RMA if you don’t feel comfortable doing it.
 
Ok thanks for all the inputs, really appreciate it. I'll try the undervolt first and assess my options from there depending on the result.

I have repasted a card before so fairly comfortable with doing that. I did it with the 2080 that I had previously and temps dropped by about 5 degrees if I recall.

Before this, I will also investigate potentially RMA the card too with Gigabyte. I only purchased the card last July, so seems a little early for paste to have degraded.
 
Justnotsleeping said:
Ok thanks for all the inputs, really appreciate it. I'll try the undervolt first and assess my options from there depending on the result.

I have repasted a card before so fairly comfortable with doing that. I did it with the 2080 that I had previously and temps dropped by about 5 degrees if I recall.

Before this, I will also investigate potentially RMA the card too with Gigabyte. I only purchased the card last July, so seems a little early for paste to have degraded.
Click to expand...
ultimately if it's unstable at stock you could argue it's a valid rma already but given it only takes a few minutes to try an undervolt there's no harm and might save the hassle of an rma. certainly though if it looked like it was still crashing after an undervolt and the gpu is determined to be the cause, i'd rma'ing before repasting. you could go to the lengths of repasting and still have the same issue but now you've potentially screwed the warranty.

anyways, get that bugger uv'd and let us know how you get on - i'm betting on it working fine (fingerscrossedemoji) :)
 
Thanks, will get to some testing tomorrow and let you know. Quick question, what do you guys use for stress testing GPU’s? I have a 3D mark license to use Timespy, etc. Are there any other good options?
 
Ok, so I undervolted the GPP from 1150 to 1050 mV, which caused crashing when running Timespy and Timespy extreme. I raised this to 1080 which seems to be more stable.

Furmark seems to have the hotspot now hitting 105C but my last run at Timespy Extreme the hotspot in HW monitor peaked at 111C, so thats a 30C variance between hotspot and standard GPU Temp at 81C.

So im not sure where I am now....seems like i should contact Gigabyte to look at RMA
 
You must log in or register to reply here.
Back
Top Bottom