Do these GPU temps look ok while gaming?

[nlel1975]

Do these GPU temps look ok while gaming?

 

[Kelt]

I'd be concerned if that was my rig.

 

[HoneyBadger]

They look fine to me, assuming you’re on air and the card was under heavy load.

 

[nlel1975]

They look fine to me, assuming you’re on air and the card was under heavy load.

Yeah VR gaming at 4K resolution.

 

[nlel1975]

Undervolted the card a bit and ran the VR 4K game for 40 minutes to test stability which it passed so might try and undervolt it even lower, here's what I have now...

 

[sgoaty]

Which 7900xtx is it?

 

[iamgigglz]

GDDR6 recommended max is 95 with throttling at 100 so you're on the bleeding edge. I'm curious to know which card it is - you might be in for some thermal tape work.

 

[nlel1975]

Its the MSI Trio version. From googling you should only be worried if it consistently goes above 105C.

 

[tamzzy]

105C

105c is the tjmax. I would be extremely worried if it went above that full stop.
98/96c is acceptable, but it isn't great though

 

[iamgigglz]

Its the MSI Trio version. From googling you should only be worried if it consistently goes above 105C.

Fair enough.

If you're feeling adventurous you could drop in some fresh thermal pads - here are the specs.

 

[Alexrose1uk]

My old RTX 3080 had similar issues, if you pushed the VRAM on it with any real load or intensity, it'd easily hit over 100 degrees, and could get close to the 110 max recommended temp; some of that GDDR6 just runs very, very warm; and I believe that's part of the reason many 9070XTs swapped from Hynix to Samsung (even if it was slightly less performant, believe the Hynix operated slightly tighter timings out the box) as the newer Samsung VRAM ran much cooler out the box (often ~20 degrees or so), and in many cases I believe it overclocked better too.

 

[nlel1975]

Well I came back to my PC this afternoon at 12:02 pm after having it set to auto shutdown at 10:54 am and it was stuck shutting down. Held the power button to switch it off and tried to boot it.

It wouldn’t load windows just got stuck with the spinning circle. So I somehow managed to get it into a recovery mode where I did a system restore from 10/06/26 which worked.

It’s back to the way it was in the adrenaline software before I undervolted it. I guess the undervolt wasn’t stable. Won’t be trying that again and it didn’t make much difference anyway.

 

[something daft already!!]

That does seem very warm. Mine runs at most, around 60C under full load. It might need a repaste. Not something I would fancy doing.

 

[Finners]

For a quick test take the side panel off the PC, if you see a decent drop in temps you know it's a fresh air getting to the GPU issue or a cooler issue.

Won't fully diagnose why you have warmer temps but points you in the right direction

 

[nlel1975]

Well I set a custom fan curve...

25c 20%
50c 30%
60c 50%
70c 75%
75c 100%

Now have these temps...



A full 10c off the memory.

 

[pooley]

That does seem very warm. Mine runs at most, around 60C under full load. It might need a repaste. Not something I would fancy doing.

Agreed, mine was 80+ until I put some Honeywell PTM on the die and now also around 60c.

 

[tamzzy]

Well I set a custom fan curve...

Now have these temps...
A full 10c off the memory.

in a nutshell, you're basically just bruteforcing ram cooling by ramping up the fans via gpu core temp
it works, but i assume at the expense of noise

 

[nlel1975]

it works, but i assume at the expense of noise

Doesn’t bother me I can’t hear any noise when wearing my VR headset, the game noise comes through the headset.

 

[nlel1975]

Here's what my fan settings looks like...



You can open the fine tuning controls and enter the numbers which is easier than trying to manipulate the graph. This is all done in the AMD Adrenalin software of course.

 

[RollerBall]

Pretty good improvement that. What case do you have out of interest?