When the plotter component is used (for graphs etc) in an "updating" manner (see, e.g., the CPU widget or the network traffic widget) there is a noticeable increase in CPU usage. The effect is cumulative: the more plotter widgets are shown, the higher the CPU usage is. This has been observed with both recent Mesa git (to be 10.6) and with Mesa 10.5. Radeon and intel drivers seem to trigger this behavior (I don't own NVIDIA hardware, so I can't test). It wasn't present when the component was first introduced: f4e2f0685a92428b5ae24360a7f6dfc937a62987 introduced it, in fact removing the update() call makes the CPU usage go down. Reproducible: Always Steps to Reproduce: 1. Add a network monitoring widget to Plasma Actual Results: CPU usage increases, proportionally with the number of widgets (plotters) shown. Expected Results: CPU usage does not increase.
i tought QWindow::update() was kindof even compressed already... in this case we should find a way to do only one queued update per window when a lot of widgets ask for it in a short time
I can confirm that bug. It is quite annoying given that the cpu usage of the plotter component makes the fan of my computer get active.
Had same issue with plasma 5.3 recently installed on Arch. With CPU Load plasmoid, high CPU usage for Xorg (~60%), fans spinning. Removing plasmoid solves the issue.
I can confirm this too, as I have mentioned here: https://forum.kde.org/viewtopic.php?f=289&t=124937&sid=8a338fe81f6c00204fb3d86f0a5fb0e7&p=334163#p334149
I also saw similar behavior in the plasma network manager widget and the telepathy systray component, is that plausible or should I investigate those further?
This occurs on Manjaro Plasma 5.3.1 (3,18 Kernel). CPU useage returns to normal after all widgets removed - CPU monitor seemed to be the worst culprit.
If this was "RESOLVEd FIXED" why is it still happening in 11-2017?
And you know the bug is this?