I have a dual-boot computer and usually use the GPU HDMI port of my NVIDIA to connect my screen; But when using Fedora the GPU fans don’t run; eventually the GPU overheats and the computer goes black and crashes. I’m very new and I’m not sure how to solve this (technically I could use the motherboard port every time I use Linux but this is an impractical and cumbersome solution).
The only time I have had that scenario was when the fans on the GPU had failed. Replacing the fans fixed it.
However, if I am understanding your post it seems that either the fans run when running windows or that there is little enough stress on the GPU that it never overheats.
Is that correct or am I misunderstanding?
Are you using the default nouveau driver for the GPU or have you installed the nvidia driver from rpmfusion? The nvidia driver usually controls the GPU fan speeds and temps.
The nouveau driver does not fully support the newer nvidia GPUs.
Please provide the output of inxi -Fzxx as preformatted text.
Yes, that’s the case; the fans are easily manageable on Windows GeForce Experience; nevertheless fans just don’t move on the Fedora Boot; even though I installed the non free RPM Nvidia Driver.
Hi - I see you edited in your system information, and in there it looks like it is actively using the nouveau driver for your NVIDIA card? I have the RPM non-free NVIDIA driver installed, and this is what my inxi -Fzxx shows in the Graphics section:
The fact that yours says driver: nouveau still would tell me that the NVIDIA drivers either aren’t fully installed or aren’t active - just to check, have you tried removing and reinstalling them using the RPMFusion guide?
As noted, this clearly shows the nouveau driver in use.
Please show us the installed nvidia drivers. dnf list installed *nvidia* will show all needed info if you installed from the rpmfusion repo.
Hmm, just as @computersavvy suspected, looks like something didn’t actually execute in the RPMFusion driver installation - here’s what dnf list installed *nvidia* on mine looks like after installing from rpmfusion:
Think you might need to try that installation again, using the RPMFusion site’s instructions (perhaps you enabled the right repository, but the install process just didn’t complete afterward?)
Don’t know why, but after the upgrade the driver would not function so I removed everything and reinstalled then it worked.
To avoid removing the firmware package I used dnf remove *nvidia* --exclude=nvidia-gpu-firmware then reinstalled with dnf install akmod-nvidia xorg-x11-drv-nvidia-cuda.
Yes, I use apps that require cuda and have for a long time.
Hi; I applied the recommended dnf remove *nvidia* --exclude=nvidia-gpu-firmware
and then reinstalled with dnf install akmod-nvidia xorg-x11-drv-nvidia-cuda.
This is the new result. Fans are not working still because inxi -Fzxx | grep fan
Glad to see that you now have the nvidia driver installed and functional.
What is the info you see with the nvidia-settings app related to fan and temp?
How do you know the fan is not running? Have you looked at it while the system is running, or just going by the reported speed in inxi? The reported speed may or may not be correct.
On mine, a GTX 1050, I do not get a fan speed reported in inxi, but the thermal temp remains near 60 C while running a GPU process from boinc using cuda. I can see the temp with lm_sensors and gkrellm as well as both fan speed and temp within the nvidia-settings app. Note that I leave the fan in auto control and it keeps the temp near where shown in the image below.
When I launch NVIDIA settings the Fan Speed is always set to 0. If I do Enable GPU Fan Settings it will start the fans immediately but it’s not the default. Meaning that every launch the fans will be off unless I set it here.
I have found some alternatives and will report results:
I see you have manually set the speeds. That is fine. I can do so as well if I choose.
The only drawback is that the fan is now locked to the speed you select so it may require more tweaking to maintain temps or running the fan at a higher speed than actually required to make sure it does not overheat with load changes.
With a single GPU the automatic control keeps the temp at about 60 C for me and I am happy with that. Note that the reported temp with nvidia-settings is for card 0 only.
Using lm_sensors and gkrellm probably will allow you to monitor temps on multiple cards at the same time.
Fedora 37 & fedora 41 are drastically different, as well as using wayland instead of X11 for the DE.
You are opening a necro thread that already had a solution posted
I am closing this one, and if you have a question that is related where you desire an answer then please open your own topic and provide details so that we may be able to assist.
