Feature Request: Add GPU temp/usage

[DamianUntapped]

Would it be possible to add support for GPU temp and usage like the CPU temp and usage?

[FearNaBoinne]

I support this request! :)

[GM-Script-Writer-62850]

This is a rather tedious request, different code for nvidia/intel/amd cards/chips, also varies kernel versions and probably different between distros
then you have old hardware that needs the old driver and the new drivers needs different code than the old one
i'v done this in a kde widget as well as xfce4-genmon-plugin before and every time it is specific to my current hardware

[FearNaBoinne]

Doesn't have to be... Just allow setting an input source, and the end-user can point it at /sys/class/hwmon/hwmon#/thermal#_input, /sys/thermalzone/* or any other source that shows it in mDeg. (You can even have a little script poll nvidia-smi for the GPU temperature into a textfile that way.)

I've done this for desklets in the past.

[DamianUntapped]

This is a rather tedious request, different code for nvidia/intel/amd cards/chips, also varies kernel versions and probably different between distros then you have old hardware that needs the old driver and the new drivers needs different code than the old one i'v done this in a kde widget as well as xfce4-genmon-plugin before and every time it is specific to my current hardware

Wouldn't it be the same for the CPU, intel/amd/qualcom?

[GM-Script-Writer-62850]

for nvidia you would need to reply on nvidia-smi as far as i know (unless you have the FOSS driver), for amd well take a look at this
https://github.com/GM-Script-Writer-62850/do-it-yourself-bar/blob/master/examples/contrib/kde.diybar.amdgpu.sh, you can't get everything from the sensors command

maybe someone knows a better way?

[FearNaBoinne]

You can create an automatic script that uses nvidia-smi to output to a textfile once per second and read the textfile, or you can let it accept the output of a command instead, and create a small script that uses nvidia-smi and grep/sed/awk to only parse the temperature
AMD typically exposes through lmsensors/hwmon, so you can either read /sys/class/hwmon/hwmonX/temp1_input or use another script to parse sensors -j
Similar for other tools.

It doesn't have to be complex to allow this to work. Just create a plugin that either takes the output of a command or reads a 'file' (hwmon are 'files' as well as far as readfile/openfile() is concerned) and displays it.
Allow custom label and unit, and you have a universal display you can use for anything that can be read from a file or stdout...

[DamianUntapped]

I used the command suggested here:
[https://github.com/StreamController/StreamController/discussions/397]

I did it as a run command action under the os plugin and set it to auto run every 1.0 seconds. Seems to be working just fine. I'd think this could be a feature added pretty easily to be like the CPU Temperature action under the os plugin. Then under the os plugin there could be an option like "GPU Temp - NVidia" and I could set it up exactly the same as the CPU Temperature action under the os plugin.