MSI Duke RTX 2080 Fans spin at max speed, ignoring custom fan curve

[Michael.Garnier]

 I've had my RTX 2080 (MSI Duke) edition since launch and it has always had this quirk of spinning the fans to 3500+rpms becoming very loud whenever the temperatures hit the mid 70s celsius (not any specific temperatures, just in that range)

The card will follow a custom fan curve that I create in MSI Afterburner up until this temperature range and higher, then it completely ignores the fan curve and slams the fans to maximum speed
The strange part is that the Fan speed % metric will read something random, yet the fan tachometer will be over 3500 rpm no matter what, this is tested by setting my fans to a static speed in MSI Afterburner and then launching a game which pushes the temperatures above the "threshold" and then the fans ramp up, though the Fan speed % stays at whatever I set it to

I've reinstalled MSI afterburner, tried every combination of "auto" and "user defined" in Afterburner
I've fully removed all video card drivers via DDU and did a fresh install of the latest nvidia drivers (461.09)
This issue has been present since I've had the card
This issue is not a matter of high temperatures, it is a matter of the fans ignoring any fan curve after it hits a certain (also quite reasonable) temperature

Any help or advice on this would be great
Scoured the internet and couldn't find any issues that were resolved that were the same as mine
TIA

 

[Plermpel]

I don't know about the MSI card but I experienced something similar on my Strix 1080Ti. When the custom curve is too flat the GPU sometimes spikes in temperature ranges that are not expected at that state of the fan curve, it's some sort of protection, the card ramps the fans until it reaches a certain temperature and then goes back to your fan curve. Does it also do that in Auto mode? Then maybe you need new paste or your cooler needs to be reseated. Also sometimes VBIOS update helps with such issues if there is one availiable.

[Michael.Garnier]

15 minutes ago, Plermpel said:

I don't know about the MSI card but I experienced something similar on my Strix 1080Ti. When the custom curve is too flat the GPU sometimes spikes in temperature ranges that are not expected at that state of the fan curve, it's some sort of protection, the card ramps the fans until it reaches a certain temperature and then goes back to your fan curve. Does it also do that in Auto mode? Then maybe you need new paste or your cooler needs to be reseated. Also sometimes VBIOS update helps with such issues if there is one availiable.

These are the fan curves i've been testing with, not sure if you'd make any alterations
the temps don't even hit 80 so i'm not convinced that this is an issue with the cooling capacity of the card
I have heard for issues that are in the same ballpark as this that there sometimes is something with the bios of the card that overrides a software fan curve but i don't understand why this would be implemented on a high end card like this meant for gaming

I could see it if this was a quadro or something, but i don't get why it would be on an geforce card

[Michael.Garnier]

ALSO, this is not my video but this is exactly the behavior mine exhibits, sound is the exact same as what i hear with what sounds like a second fan kicking in or something
The temps in this video (69-70C) are slightly lower than what my temps are when this happens but i'm assuming it is the same root cause, whatever it may be

[Plermpel]

I would just try to update the VBIOS if that's an option. I could also imagine that the card looses track of the tacho signal for a very short time and goes into some sort of save mode but I remember my Strix 1080Ti doing the same sometimes it didn't crank the fans like this but switched to the stock fan curve until it hit 50°C  again.

[vervurax]

Hi, I'm the dude from that video and I did solve this problem!

All I had to do is apply fresh thermal paste on the GPU. Didn't think it would help, since the card isn't that old but it made a world of difference.

Here are some pics so you know what to expect: https://imgur.com/a/VSEfiM7

 

There's a lot of guides on how to do it so I won't write another one, but let me know if you want some pointers. It's pretty simple if you know how to use a screwdriver  Just be gentle, don't lose any of the thermal pads and prepare some alcohol for cleaning (ideally IPA).

 

As for fan curves I reverted everything back to default and let the card's BIOS do the work. It's not a software issue, that's why Afterburner wouldn't help. I think the thermal paste quality or the way it's applied from factory isn't very good, which creates hotspots on the die and sensors go crazy. If it's anything like CPUs then the highest temperature reading is considered, not average. So as long as you have an even layer of good paste everything should be fine. 

[Michael.Garnier]

43 minutes ago, vervurax said:

Hi, I'm the dude from that video and I did solve this problem!

All I had to do is apply fresh thermal paste on the GPU. Didn't think it would help, since the card isn't that old but it made a world of difference.

Here are some pics so you know what to expect: https://imgur.com/a/VSEfiM7

 

There's a lot of guides on how to do it so I won't write another one, but let me know if you want some pointers. It's pretty simple if you know how to use a screwdriver  Just be gentle, don't lose any of the thermal pads and prepare some alcohol for cleaning (ideally IPA).

 

As for fan curves I reverted everything back to default and let the card's BIOS do the work. It's not a software issue, that's why Afterburner wouldn't help. I think the thermal paste quality or the way it's applied from factory isn't very good, which creates hotspots on the die and sensors go crazy. If it's anything like CPUs then the highest temperature reading is considered, not average. So as long as you have an even layer of good paste everything should be fine. 

Sounds good, thanks a million for the reply btw

I replaced the thermal paste on my XFX R9 390 a few years back, i'm assuming it will be a similar procedure
I'll try this and let you know
I have read in a few forums that some of the msi cards have had poorly applied and/or cheap paste so i'll throw some   mx-4 on it and see how it goes

 

[Michael.Garnier]

So finished re-applying the thermal paste
It does seem as though i can use whatever fan curve i want now, so that issue is fixed
The downside is that my temps cap out about 10C hotter than they did before
After about 30 minutes of testing the GPU temps sits at 87C which should be fine, i ran my r9 390 at 94C for years and the 2080 has a max safe temp of 88C so it should be alright
Thanks for the suggestion, if you have any idea why my temps are so much higher though let me know and I can give it a shot

[Michael.Garnier]

Update: I just seen that only one of my gpu fans are spinning, the other 2 won't spin
This may be the reason why temps are higher, going to see why this is the case

 

 

EDIT: Cannot see why 2/3 fans aren't spinning
Idle temp is very high, around 58-60C
Maxing at around 88C under load
Only the fan closest to the I/O plate is spinning
both the fan headers are fully plugged in, as well as what i'm assuming is the rgb cable

I'm at a loss

[Mark Kaine]

47 minutes ago, Michael.Garnier said:

Update: I just seen that only one of my gpu fans are spinning, the other 2 won't spin
This may be the reason why temps are higher, going to see why this is the case

 

 

EDIT: Cannot see why 2/3 fans aren't spinning
Idle temp is very high, around 58-60C
Maxing at around 88C under load
Only the fan closest to the I/O plate is spinning
both the fan headers are fully plugged in, as well as what i'm assuming is the rgb cable

I'm at a loss

Oof, well maybe you plugged in something wrong - idk... 

 

I would suggest posting pics of the card, the fan headers, everything... hopefully someone sees something (probably not me though as I've never opened a gpu so I wouldn't know) 

 

but something... must be up, I doubt it's software related, but who knows.

 

[Michael.Garnier]

So I took the card out once again and reseated the fan headers
I also unplugged the RGB header as well but could not reseat it since I must have overtightened one of the backplate screws and it's spinning the threaded rivet around when i try to unscrew it so the card is permanently assembled now lol

However after reseating the fan headers and leaving the rgb header unplugged, i plugged the card back in and all 3 fans are working perfectly now, my temps are about the same as what they were before i did anything to it (idling in low 40s, around 78C max under load) however the fans ramping up to 3500rpms has stopped
I think i count this as a win

Thanks for all the help, marking this one as resolved

[SavageNeo]

3 minutes ago, Michael.Garnier said:

So I took the card out once again and reseated the fan headers
I also unplugged the RGB header as well but could not reseat it since I must have overtightened one of the backplate screws and it's spinning the threaded rivet around when i try to unscrew it so the card is permanently assembled now lol

However after reseating the fan headers and leaving the rgb header unplugged, i plugged the card back in and all 3 fans are working perfectly now, my temps are about the same as what they were before i did anything to it (idling in low 40s, around 78C max under load) however the fans ramping up to 3500rpms has stopped
I think i count this as a win

Thanks for all the help, marking this one as resolved

I do not see why the thermalpaste would have been the problem. IF anybody is having similar problems I may rederiect them to this thread. 

 

Also you can use the Checkmark button on the  @vervurax' comment