Fix gpu usage scaling or inf values in plasma-systemmonitor graphs; also fix nvidia-smi process leak

Background On my system (with intel and nvidia gpu) the gpu usage an memory graphs shown in plasma-systemmonitor where completely broken. See attached screenshot: gpu_usage_memory_broken

With this change graphs now display useful values 😄 gpu_usage_memory_fixed I compared the vram values of processes with nvidia-smi and intel_gpu_top, they look plausible now.

In addition the nvidia-smi call happened in a loop for every line read which is wasteful.

Edited by Bernhard Friedreich

Merge request reports

Loading