vulcZ
Not sure if this is related, but in today's driver release notes there is a current known Windows 10 issue listed that might be what we've been experiencing...

 
Taken directly from .
 

Sajin
Here are the top causes of why a GPU would run at elevated clocks at idle...

#1 You're running multiple monitors. Fix: Enable mutil display power saver using nvidia inspector.
#2 Your machine has a virus. Fix: Scan computer using malwarebytes.
#3 One or more programs running in the background are using your GPU. Fix: Perform a clean boot of windows.
#4 Power management mode inside the nvidia control panel is set to prefer maximum performance. Fix: Set power management mode to adaptive then reboot your computer.
#5 Running a resolution of 2560x1440 @ 144Hz. Fix: Change refresh rate to 120Hz.
#6 K-Boost is enabled inside Precision X. Fix: Disable K-Boost.
#7 Shadowplay is enabled inside geforce experience. Fix: Disable shadowplay inside geforce experience OR completely uninstall geforce experience from your machine (highly recommended)

colma11
I'm glad I found this thread.
I've got a Gigabyte RTX 3080 Eagle OC, and it basically started acting weird overnight.
The clock got stuck at 1755mhz which was causing the fans to rev up really loudly every twenty seconds to cool it back below the treshold.
It also started to coil whine every now and then, even when idle. In games it's quite annoying. I fixed the clock issue by setting Nvidia power management mode to optimal power and restarting my PC.
The day before that happened I was downloading some weird **** like Docker and Linux Subsystem for Windows that I need for my work, so my first idea was that these where the culprits, but it's apparently not the case.
 
My clock is now unstuck but the noticeable coil whine persits which got me thinking that perhaps changing the power mode only covers-up the a deeper issue. What do you guys think?
 
I can recreate the issue and lock the fan speeds @ 1755mhz by setting power management mode back to max performance. Which is weird because it had been running fine at that setting for weeks until yesterday.
 
edit: I feel like there must be something running in the background that's requesting the clock to run at 1755mhz, because that'd be the only reason why it happened overnight. There could be a priority issue where one hidden process forces it at 1755mhz and nvidia power management being able to overwrite that. I tried clean booting with most process/services off, but if I set it at max performance, it still locks up at 1755mhz. hmm...
 
edit2: read somewhere that these could be a GPU fan issue. Come to think of it, HWInfo only displays GPUFan1 and GPUFan2 readings, even though mine has three fans.
 
edit3: I might have finally tracked down the issue. With prefer maximum power mode and the clock being stuck at 1755Mhz I started to just kill the 'Start' process repeatedly in task manager. Most of the time it reappeared, but I did it over and over again until the clock dropped back to 210Mhz for the first time in a couple of days with the 'prefer maximum power' mode on. After a while ,however, it locked back up to 1755Mhz, probably because Start kicked in again. So I went over to NVCP and just went through every program that had something to do with Windows: Photo Viewer, Skype for Business, Camera (which I don't own), Windows Lock Screen, Metro Video Player, Minecraft, MS Store, SearchUI, Microsoft PC Settings, MS What's New, Gaming Overlay and set it all to Optimal. Now the system boots up all right at 210Mhz, which is what I like to see, but whenever I open up the start menu it will immediately spikes up to 1755Mhz. Must be some Windows issue, perhaps a corrupted Windows update that happened overnight? Or could it be some Windows mail stuff hapenning in the background? Will check tomorrow.
 
And since some people say it has something to do with the display you're running: mine is LG27GL850 @ 2560x1440, 144hz.

colma11

Your "EDIT 3" makes sense... but darn, that is a lot of services and apps running in the background that could cause this. Do we know if this same issue is happening accross the board?  I feel like it's either a firmware or driver issue. I recently saw a video from BPS customs OC'ing the XC3 Ultra where he showed PX1 with clocks at 210Mhz when system was idle. That was at release date, so surely had the release BIOS and driver. 
 
Has anyone tried flashing to a previous BIOS version, previous Driver, or Both?

I did some further testing, and honestly, I'm more confused than I was before. My PC boots up with 210Mhz clock with "maximum performance mode" on, but using pretty much any program that has practically 0 load on the GPU, will cause the clock to run at 1755Mhz. Like: Visual Studio Code (which is basically a text editor), Windows mail app, MS Teams even mouse software GUI.
 
At this stage I decided I'd stick to to the optimal/adaptive setting and just put 'max performance' on games only. This way it works alright, but I'm still worried I could run global max performance with no problems for weeks til recently.
 
edit: I found that if I do NVCP -> Desktop -> Display GPU Activity Icon in notification area, open up the tray icon that shows everything that has impact on GPU and then close everything what's listed (apart from basic Windows functionality like searchapp.exe or textinput.exe), it's the only way I can bring it down to 210Mhz. So it seems like pretty much anything that uses GPU automatically forces it to run at boost clock (1755Mhz in my case) and overheat. My understading is, and you guys probably can confirm, that 'prefer maximum power' should not force the clock to run at boost at all times; instead, it should downclock accordingly. Otherwise, and that's the case with my card right now, using basic text editor like VSCode causes the card to overheat and draw 100W of power. I no longer think that this has anything to do with Windows. My suspicion now is either nvidia driver (but...just like that? Ceased to work correctly overnight?) or just hardware issue, especially that I had some issues with it the moment I got it - the pins on my RTX 3080 sat too low to be able to connect 8pins (common problem), and I had to tear the card down and push the wires forward.
 

colma11

selarom
I must also add to your findings. My previous post is basically null and what I have found is exactly the same as you... Setting up adaptive power does fix the issue and now I have to hunt down the executables to make all my games run on High Performance. That sucks! I hope everyone else has the same behavior so we can call this an identified issue..

Hi,
 
are you running i7-10700K CPU by any chance? I'm also posting on overclockers forum, and some user who's having the same problem where the GPU won't downclock correctly, is also running that CPU. So maaaaaybe that's related somehow.
 
 
 
 

bradleynowell96
I'm having problems too, but it looks like the EVGA cards just don't keep cool as others. First time I've ever regretted buying one. I had a Gigabyte 3080 that stayed at around 68 degrees under load with normal fans and an good overclock. My FTW3 on the other hand without an overclock AND fans set to 100% gets to high 70s/early 80s and just continues to downclock and become hotter. Ridiculous in my opinion. Contacted EVGA they just said it's "normal" for this card. Wish I knew beforehand I would have kept my Gigabyte.

Omoeba
This is normal when Nvidia Broadcast / RTX Voice is running