[SOLVED] GPU temperature limit issue

Jul 20, 2021
5
1
10
Hello guys, I am using a MSI VENTUS 3090. The temp limit in MSI afterburner is set at 83c , but in actual the temp keep under 70c, as soon as it hit 70c, the clock reduce to like 1400 or 1500 . how can I break that 70c limit.
 
Solution
Could be a coincidence. Something else on the card may be thermal throttling, as power throttling doesn't incur as large a penalty on the frequency.
Hitting power limits will make the card throttle back, but the drop is usually minor, as I just mentioned.
Optimal power in Nvidia Control Panel throttles back core clock when the gpu isn't being used much. It's the default power plan. Loading screens and some menus don't need the gpu to run 1900mhz+.
Might want to check Vram temperature(hwinfo), but I don't think memory thermal throttling should force down gpu core clock...

Did you alreadly uninstall the previous gpu's(if there was one) drivers and then install the one for the 3090?
Display Driver Uninstaller, NVClean Install, and Geforce...

Phaaze88

Titan
Ambassador
Could be a coincidence. Something else on the card may be thermal throttling, as power throttling doesn't incur as large a penalty on the frequency.
Hitting power limits will make the card throttle back, but the drop is usually minor, as I just mentioned.
Optimal power in Nvidia Control Panel throttles back core clock when the gpu isn't being used much. It's the default power plan. Loading screens and some menus don't need the gpu to run 1900mhz+.
Might want to check Vram temperature(hwinfo), but I don't think memory thermal throttling should force down gpu core clock...

Did you alreadly uninstall the previous gpu's(if there was one) drivers and then install the one for the 3090?
Display Driver Uninstaller, NVClean Install, and Geforce Experience can all do it... I've never tried the last 2 though.


You could also have a faulty card, such as stuck at low frequency due to short circuit - but lets shove that aside and hope that's not the issue.
 
Solution
Jul 20, 2021
5
1
10
Could be a coincidence. Something else on the card may be thermal throttling, as power throttling doesn't incur as large a penalty on the frequency.
Hitting power limits will make the card throttle back, but the drop is usually minor, as I just mentioned.
Optimal power in Nvidia Control Panel throttles back core clock when the gpu isn't being used much. It's the default power plan. Loading screens and some menus don't need the gpu to run 1900mhz+.
Might want to check Vram temperature(hwinfo), but I don't think memory thermal throttling should force down gpu core clock...

Did you alreadly uninstall the previous gpu's(if there was one) drivers and then install the one for the 3090?
Display Driver Uninstaller, NVClean Install, and Geforce Experience can all do it... I've never tried the last 2 though.


You could also have a faulty card, such as stuck at low frequency due to short circuit - but lets shove that aside and hope that's not the issue.
The card runs pretty well in the past 6 months. And I check my previous 3DMARK result, the card can reach 78 to 80c. Ive updated nvidia driver, reinstall MSI afterburner but not helping.
 

Phaaze88

Titan
Ambassador
The card runs pretty well in the past 6 months.
Ok, did you remove/reinstall the card at any time after that?

the card can reach 78 to 80c.
That's core, or Vram? I was asking about the Vram.
That result is fine for core temperature.

Ive updated nvidia driver
Afterwards, NVCP power plan was set to Adaptive or Max? Updating the driver defaults it to Optimal, which the driver tries to save power whenever possible.
 
Jul 20, 2021
5
1
10
Ok, did you remove/reinstall the card at any time after that?


That's core, or Vram? I was asking about the Vram.
That result is fine for core temperature.


Afterwards, NVCP power plan was set to Adaptive or Max? Updating the driver defaults it to Optimal, which the driver tries to save power whenever possible.
I didnt touch the card. I just run a three kingdom total war benchmark and check the gpuz:the gpu temp keep under 70c, vram tempture is also 70c, hot spot temp is around 105c.
 

Phaaze88

Titan
Ambassador
Ok. Card wasn't touch recently.
What about NVCP power plan then? Did you take that off Optimal?

That hot spot isn't great though. A 35C gap between that and the core is pretty high. A more typical gap would've been about 20C.
Card starting to sag, or the paste has dried up somewhere?
 
Jul 20, 2021
5
1
10
Ok. Card wasn't touch recently.
What about NVCP power plan then? Did you take that off Optimal?

That hot spot isn't great though. A 35C gap between that and the core is pretty high. A more typical gap would've been about 20C.
Card starting to sag, or the paste has dried up somewhere?
power plan is set to maximum performance, not working neither. Maybe I should clean the card.
Thank you!
 

Phaaze88

Titan
Ambassador
power plan is set to maximum performance, not working neither. Maybe I should clean the card.
Well, not so much as clean it, but...
-you haven't had it that long.
-you haven't touched it since you installed it.
-it was working just fine during the short time you've owned it, now it throttles hard.

So I'm thinking, either the paste has dried up, or the card has fallen victim to sag already.
When sag happens, the heatsink shifts, and some areas don't make contact as well as they used to.
 
Jul 20, 2021
5
1
10
Well, not so much as clean it, but...
-you haven't had it that long.
-you haven't touched it since you installed it.
-it was working just fine during the short time you've owned it, now it throttles hard.

So I'm thinking, either the paste has dried up, or the card has fallen victim to sag already.
When sag happens, the heatsink shifts, and some areas don't make contact as well as they used to.
I bought a 3080ti, let the 3090 rip.
 
Last edited by a moderator:
  • Like
Reactions: bombo007