Hello,
I've recently begun having some issues with my RTX 3080 which until recently had been working fine. I booted my PC one day and noticed the fans revving up very loud every twenty seconds. So I opened HWInfo to see what's going on, and apparently the clock was stuck at boost speed (1755Mhz) which caused the card to overheat rapidly, so the fans revved up at an interval to cool it down below 55 degrees celsius threshold. My GPU load at that time was 0-1% which made it all the more weird. High clock at boost speed obviously causes a 100W of power draw which is aweful.
Apart from that, I began noticing significant coil whine (which sometimes even kicks in when idle) that wasn't there before. In games it's loud and quite annoying.I later found out that if I open up Nvidia Control Panel and set the power management mode to Optimal or Adaptive (instead of prefer max performance), the clock will go down to 210Mhz which is what I like to see. However, for weeks till recently, I had been running global max performance setting with no issues whatsoever. My understing is, and I confirmed it with some other nvidia users, that 'prefer max performance' should still downclock the card accordingly, and it happened in my case before, but now it will just lock the card at boost speed.
So I did some further testing and found out that if I shut all the running apps, the clock WILL go down 210Mhz on maximum performance mode, but opening pretty much any application that has some minimal impact on the GPU (like Visual Studio Code which is basically a text editor, Windows mail app, etc.) will instantly lock the card at 1755Mhz, draw 100W of power and make it overheat. This didn't happen before, and I've confirmed with others it should not happen. The cards temperatures are all right, 35-40 degrees celsius when I'm literally not using anything (no mail app, no nothing). Any app like those I mentioned before, will cause the card to go to 55 degrees (would go higher, but the fans rev up) and sit at 1755Mhz clock when there's practically no reason to draw so much power for such a menial task.So that's that.
I could technically run the card with Optimal/Adaptive setting, but I'm worried there is something wrong with the card that can potentially shorten its lifespan, especially that it began overnight. The coil whine, however, is present even on Optimal/Adaptive setting (when idle, 0% load). I contacted Nvidia Tech Support, and pretty much the only thing we managed to discover together is that running NVCP Debug mode can clock it down to 1710Mhz instead of 1755Mhz, but later the Nvidia tech told me to contact the manufacturer and keep him updated.
So I was wondering whether you're familiar with that issue and if maybe you'd have any suggestions as to what should I do to make it work as good as before. Some people have been reporting a similar problem on EVGA Forums, with no apparent cause, which could perhaps point to some driver issue? I'm not sure about that, however, since I did a clean uninstall and that does not explain the fact that it began overnight. Maybe some issues with the Windows 10 driver that might have been installed when I shut down my PC? I have not personally overclocked the card further.
Some of the things I've tried:
1) DDU, clean uninstall of drivers in safe mode.
2) Process Explorer => Killing pretty much any proces that could have some sneaky impact on the GPU; no matter what I do, whenever I open a basic app, the clock will lock at 1755Mhz.
3) Nvidia Inspector App can artificially limit the clock to 420 (Multidisplay power saver), but that's no solution really, only an artificial workaround/gimmick.
4) Malwarebytes scan.
None of these really can bring the card to how it was before.
I'd appreciate any suggestions.
Have a nice day.
I've recently begun having some issues with my RTX 3080 which until recently had been working fine. I booted my PC one day and noticed the fans revving up very loud every twenty seconds. So I opened HWInfo to see what's going on, and apparently the clock was stuck at boost speed (1755Mhz) which caused the card to overheat rapidly, so the fans revved up at an interval to cool it down below 55 degrees celsius threshold. My GPU load at that time was 0-1% which made it all the more weird. High clock at boost speed obviously causes a 100W of power draw which is aweful.
Apart from that, I began noticing significant coil whine (which sometimes even kicks in when idle) that wasn't there before. In games it's loud and quite annoying.I later found out that if I open up Nvidia Control Panel and set the power management mode to Optimal or Adaptive (instead of prefer max performance), the clock will go down to 210Mhz which is what I like to see. However, for weeks till recently, I had been running global max performance setting with no issues whatsoever. My understing is, and I confirmed it with some other nvidia users, that 'prefer max performance' should still downclock the card accordingly, and it happened in my case before, but now it will just lock the card at boost speed.
So I did some further testing and found out that if I shut all the running apps, the clock WILL go down 210Mhz on maximum performance mode, but opening pretty much any application that has some minimal impact on the GPU (like Visual Studio Code which is basically a text editor, Windows mail app, etc.) will instantly lock the card at 1755Mhz, draw 100W of power and make it overheat. This didn't happen before, and I've confirmed with others it should not happen. The cards temperatures are all right, 35-40 degrees celsius when I'm literally not using anything (no mail app, no nothing). Any app like those I mentioned before, will cause the card to go to 55 degrees (would go higher, but the fans rev up) and sit at 1755Mhz clock when there's practically no reason to draw so much power for such a menial task.So that's that.
I could technically run the card with Optimal/Adaptive setting, but I'm worried there is something wrong with the card that can potentially shorten its lifespan, especially that it began overnight. The coil whine, however, is present even on Optimal/Adaptive setting (when idle, 0% load). I contacted Nvidia Tech Support, and pretty much the only thing we managed to discover together is that running NVCP Debug mode can clock it down to 1710Mhz instead of 1755Mhz, but later the Nvidia tech told me to contact the manufacturer and keep him updated.
So I was wondering whether you're familiar with that issue and if maybe you'd have any suggestions as to what should I do to make it work as good as before. Some people have been reporting a similar problem on EVGA Forums, with no apparent cause, which could perhaps point to some driver issue? I'm not sure about that, however, since I did a clean uninstall and that does not explain the fact that it began overnight. Maybe some issues with the Windows 10 driver that might have been installed when I shut down my PC? I have not personally overclocked the card further.
Some of the things I've tried:
1) DDU, clean uninstall of drivers in safe mode.
2) Process Explorer => Killing pretty much any proces that could have some sneaky impact on the GPU; no matter what I do, whenever I open a basic app, the clock will lock at 1755Mhz.
3) Nvidia Inspector App can artificially limit the clock to 420 (Multidisplay power saver), but that's no solution really, only an artificial workaround/gimmick.
4) Malwarebytes scan.
None of these really can bring the card to how it was before.
I'd appreciate any suggestions.
Have a nice day.
