I’ve got a 1050Ti and I’m running Ubuntu 24.04.4 LTS. I’ve run with various different driver versions (directly downloaded from nvidia.com and also the packages provided by Ubuntu) including the latest for my card (580.159.04 - card is not supported on newer branches).
After a recent batch of Ubuntu software updates (I didn’t take note of what it was unfortunately), I’ve been having trouble with the fan on my 1050Ti. The fan starts off, as it should. When under load, it turns on and gradually increases. It continues to increase even as the temperature is coming down. If the temperature drops fast enough, the fan will stop increasing speed, but will remain constant. Often it gets all the way at or near 100% however, and just stays there until reboot. I turned on Coolbits and I’m able to turn the fan on manually from nvidia-settings as well, but setting it back to 0 just leaves the fan spinning at the speed it was already going. If I set the fan speed to 1%, the speed starts slowly and gradually increases, eventually all the way to 100%. If I set the fan to something larger like 30%, it reaches 100% very quickly and stays there. Basically the driver can always INCREASE the fan speed, but never decrease it.
Also, the nvidia-settings app and nvidia-smi both always report the fan RPM is 0, even when it is at 100%. Third party tools like GreenWithEnvy also report 0 RPM, and exhibit the same type of control… can increase the speed but never decrease.
I’ve tried uninstalling the .run driver archive, installing several recent versions freshly, including the one that was working without this problem about a week ago (580.142). I’ve tried several Ubuntu packaged versions of the driver. I’ve removed and purged them each time to switch back. They all exhibit the same behavior. Under load, fan ramps to 100% and stays there regardless of settings. I tried nouveau and it did not have this issue.
My particular concern is that the drivers themselves don’t seem to think the fan is running, reporting 0 RPM always. I think when the drivers decide to turn the fan on, they check to see the fan RPM, and if it didn’t reach the target speed, they continue to tell the fans to turn faster. Since they always think it is 0 RPM, the drivers continuously tell the fans to go faster until they reach 100%.
How can I debug this issue? I’m positive there’s some leftover junk configuration file or something causing this issue because this exact driver worked perfectly for me last week. I’m not afraid to roll up my sleeves and try some low-level stuff here, but I have no idea where to begin. Any help is appreciated.