You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
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
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.)
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?
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...
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.
+1 from my side. really wish the feature exists. i managed to show a usage-% of my nvidia gpu every second, but it would be much nicer if there were also a graph like for RAM and CPU
Activity
FearNaBoinne commentedon Sep 1, 2025
I support this request! :)
GM-Script-Writer-62850 commentedon Nov 16, 2025
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 commentedon Nov 16, 2025
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 commentedon Nov 16, 2025
Wouldn't it be the same for the CPU, intel/amd/qualcom?
GM-Script-Writer-62850 commentedon Nov 16, 2025
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
sensorscommandmaybe someone knows a better way?
FearNaBoinne commentedon Nov 21, 2025
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_inputor use another script to parsesensors -jSimilar 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 commentedon Dec 1, 2025
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.
alexsperlingz commentedon Feb 21, 2026
+1 from my side. really wish the feature exists. i managed to show a usage-% of my nvidia gpu every second, but it would be much nicer if there were also a graph like for RAM and CPU