-
-
Save frantic1048/41f56fd6328fa83ce6ad5acb3a4c0336 to your computer and use it in GitHub Desktop.
#!/usr/bin/perl -w | |
# act as a KSysGuard sensor | |
# provides NVIDIA GPU info via `nvidia-settings` | |
# Usage (e.g. add gpu temperature sensor) | |
# 1. save this file, make sure it has a exec permission | |
# 2. in KSysGuard's menu, open `File` -> `Monitor Remote Machine` | |
# 3.1 in new dialog, type `Host` whatever you want | |
# 3.2 set `Connection Type` to `Custom command` | |
# 3.3 set `Command` like `path/to/this-sensor.pl` | |
# 4. click `OK`, now you can find new sensor named `gpu_temp` | |
# which is provides GPU temperature | |
# See Also | |
# https://techbase.kde.org/Development/Tutorials/Sensors | |
$|=1; | |
print "ksysguardd 1.2.0\n"; | |
print "ksysguardd> "; | |
while(<>){ | |
if(/monitors/){ | |
print "gpu_temp\tinteger\n"; | |
print "gpu_graphics\tinteger\n"; | |
print "gpu_memory\tinteger\n"; | |
print "gpu_video_engine\tinteger\n"; | |
} | |
if(/gpu_temp/){ | |
if(/\?/){ | |
print "GPU Temp\t0\t0\n"; | |
}else{ | |
print `nvidia-settings -tq gpucoretemp | head -n1`; | |
} | |
} | |
if(/gpu_graphics/){ | |
if(/\?/){ | |
print "GPU\t0\t0\n"; | |
}else{ | |
print `nvidia-settings -tq [gpu:0]/GPUUtilization | awk -F"," '{print(substr(\$1,index(\$1,"=")+1))}'`; | |
} | |
} | |
if(/gpu_memory/){ | |
if(/\?/){ | |
print "GPU Memory\t0\t0\n"; | |
}else{ | |
print `nvidia-settings -tq [gpu:0]/GPUUtilization | awk -F"," '{print(substr(\$2,index(\$2,"=")+1))}'`; | |
} | |
} | |
if(/gpu_video_engine/){ | |
if(/\?/){ | |
print "Video Engine\t0\t0\n"; | |
}else{ | |
print `nvidia-settings -tq [gpu:0]/GPUUtilization | awk -F"," '{print(substr(\$3,index(\$3,"=")+1))}'`; | |
} | |
} | |
print "ksysguardd> "; | |
} |
@BryanByteZero , To make it work on AMD, you just need to find a way to get GPU information via command line, and extract the info you want (e.g. temperature), then replace the command in Line 31
Another way, if the sensor can be detected by lm_sensors
, then you totally do not need this external script. KSysGuard has already got it. You can manually run sudo sensors-detect
to find if there is any more usable sensor. For more about lm_sensors
, you can take a look at https://wiki.archlinux.org/index.php/Lm_sensors#Setup
I'm not familiar with AMD command line tools. Good luck~
Where do you add that code?
I've modified your script to add units and detect the amount of GPU memory.
https://gist.github.com/hacker1024/c01a773f50769bd8216fa01ea0a1ef33
Fantastic, exactly what I was looking for. I'm using @hacker1024's modified version.
Two things should be added to the comments:
- In KSysGuard,
File -> Monitor Remote Machine
is only available on custom/new tabs, but not on the default tabs, i.e. Process Table, System Load. Took me a while to figure that out. nvidia-settings -tq all
ornvidia-settings -q all
may be used to list all queryable attributes in case one wants to add something
Many thanks!
I ported your script to Bash, extended it and significantly lowered its CPU footprint:
https://gist.github.com/fonic/8f38e5e3ce5c8693ae3a23aa1af21fb9
Is possible to this on AMD opensource drivers?