After installing Nvidia driver for my MX 150 I’ve noticed the fan running constantly even when the temperature drops to about 47*C.
When I switch back to nouveau driver the fan works automatically again but there are bugs and random crashes.
I am running Fedora 30 with recent kernel (5.0)
It’s a single fan design, which is controlled by (system)bios, not the nvidia driver.
Please see this how to set up your own fan control:
[url]https://medium.com/@kasunsiyambalapitiya/fixing-constantly-running-fans-on-dell-laptops-running-ubuntu-16-04-8cf6595381d9[/url]
Thank you for the quick answer, I will definitively try this out in the next days. One thing makes me wonder though, why the fan behaves correctly with the open source driver? Is it because of different usage of the GPU or because of the nvidia driver is messing up the fan control?
When nouveau is active, the Nvidia gpu is off per default so it will produce no heat. When installing the proprietary driver, the Nvidia gpu will be on per default and drive the desktop so it will produce more heat and the fans are constantly running. The driver itself does not know anything about the fan and also doesn’t mess with it. It’s plain stupid bios control.
Could you please elaborate more on that. Does that mean that nouveau uses GPU only when needed (like games or browser) and disables it when it isn’t needed or rather it means that I have to explicitly configure nouveau to use my GPU?
Definitely agree with you
Nouveau supports PRIME offload, meaning anything will be rendered on the intel gpu unless you explicitly set an application to use the nvidia gpu by using the env variable DRI_PRIME=1:
intel:
glxinfo | grep OpenGL
nouveau:
DRI_PRIME=1 glxinfo | grep OpenGL
The proprietary driver only supports PRIME output, meaning everything will be rendered on the nvidia gpu.
Furthermore, nouveau doesn’t support reclocking so it will always run on minimum clocks while the nvidia driver will ramp up on usage and produce more heat.
I wasn’t aware that my laptop has Intel card onboard, but when I run the code that @generix suggested I see Intel:
# DRI_PRIME=1 glxinfo | grep OpenGL
OpenGL vendor string: Intel Open Source Technology Center
OpenGL renderer string: Mesa DRI Intel(R) HD Graphics (Whiskey Lake 3x8 GT2)
Here is lspsci result:
lspci -k | grep -i vga
00:02.0 VGA compatible controller: Intel Corporation UHD Graphics 620 (Whiskey Lake)
lspci -k | grep -i 3d
01:00.0 3D controller: NVIDIA Corporation GP108M [GeForce MX150] (rev a1)
It’s an Optimus i.e. hybrid graphics notebook meaning the integrated intel igpu is connected to the display and the nvidia discrete dgpu has no outputs. I don’t know about the state of autoconfiguring this in Fedora, maybe read up on those:
[url]https://rpmfusion.org/Howto/Optimus[/url]
[url]https://github.com/wildtruc/nvidia-prime-select[/url]
[url]https://devtalk.nvidia.com/default/topic/1022670/linux/official-driver-384-59-with-geforce-1050m-doesn-t-work-on-opensuse-tumbleweed-kde/post/5203910/#5203910[/url]
There’s also bumblebee.