Hello,
I know this is an old thread but like many others I have recently come across this issue with my FTW3 EVGA 1080ti. (to be clear this issue seems to affect multiple generations of GTX card...I've seen as far back as the 600 series.) Actually, I've been dealing with it for almost a year now. I've always had a multi-monitor setup ever since I got this card. From the first day I put the tower together, I had idle temperatures around 54 degrees celcius and boosted clock speeds to the maximum. initally I had a 2 monitor setup, but eventually moved to a 3 monitor setup. it's important to note here that my 2 monitor setup also suffered from the locked in raging clock speeds.
After about a year I got tired of seeing such high numbers, after talking to some other owners of 1080tis who's idle temps were far nicer than my own, so i started digging. I came across this thread and gave nvidia Inspector a shot. Unfortunately for me, while it did lower my idle temps, it also came with drastic fluctuations in GPU usage associate with simple mouse and keyboard cursor movement that I couldn't understand, and didn't want to deal with...wasn't really sure if it was affecting performance or anything but it just didn't seem like you'd want going on.
So that's when i got into it with a nVidia tech support rep. Of course we can all come to the conclusion that the rep didn't really provide any real solutions and eventually just blamed the issue on some third party and gave up. I did however through all the troubleshooting I did for the tech narrow down the possible reasons why this driver error, as I believe it is, occurs.
Something I noticed, the clock speeds would only lock up at maximum if I have an HDMI + Display port cable connected to a display via the 1080ti outputs. With that combination of outputs, the 1080ti locks at maximum clock speeds no matter the resolution or refresh rate of either monitor, as far as I can tell. I was able to ad a second display, connected directly to the 1080ti using a DVI cable, and still the proper down clocking for the GPU would be in effect, temps remained very low at idle.
So in my case at least HDMI + Display Port was causing the issue...my best guess is that something is wrong with he driver and it prevents the card from down clocking for some reason. DVI+ Display Port as well as HDMI + DVI work fine for me, and do not lock up the clock speeds. It doesn't seem to matter how many monitors you have, i think it's just a coincidence that most people moving up to 3 monitors will inevitably be using a combination of HDMI and Display Port cables. 1. Because you can't connect more than 1 of anything but Display port cables. and 2. because HDMI carries uncompressed 11.2 audio...no other cable format can do that which is a huge reason why I use HDMI in my setup.
My solution was to move my Display Port, and DVI connections (even though I didn't have to move the DVI) to the onboard graphics. I don't game on the secondary monitors and the onboard GPU still renders 4k 60Hz so it works out fine for me, who has the onboard options.
I know didn't provide much of a solution , and hopefully most people are fine with nvidia Inspector. For the few in my shoes who had problems with nvidia Inspector, or simply do not want any avoidable third party software, simply making sure you do not have an HDMI and Display port display connected directly to your GTX graphics card you should be able to achieve down clocking just fine. Moving either HDMI, or Display port to the onboard graphics unit should solve your problem.