Created attachment 136423 [details] Comparison of the old and new graphs with different settings and sizes SUMMARY There are multiple issues here. In the default size, the percentages below the CPU graph are unreadable. This should be partially fixed by https://invent.kde.org/frameworks/kquickcharts/-/merge_requests/28 But KSysguard solved that more nicely, by first shrinking the legend to just colored percentages. You should still wrap them at some point, because otherwise 48 threads will fit on pretty much no screen, but having a colored line next to it is not very useful at that size. Secondly, the default of using a stacked chart leads to a chart going to 1600%, but because only one core is pinned, that one sits at 1/16th of the chart. This does not provide much useful information about why an application hangs, since it looks like it is hanging in IO instead of in a CPU loop. Third, there is no actual unstacked line graph mode. The only option is a line mode with everything below it filled, which is very hard to read and to recognize patterns in (i.e. everything spiked to 100% but with some drops here and there to 0). KSysguard solved that by having the area below the line half transparent, which allowed you to see all the lines in any case. STEPS TO REPRODUCE 1. Open CPU widget on a system with 16 cores or more OBSERVED RESULT Badly scaled CPU graph, that hides half the information. EXPECTED RESULT CPU graph that shows information of all CPU cores at a glance. SOFTWARE/OS VERSIONS Linux/KDE Plasma: (available in About System) KDE Plasma Version: 5.21.2 KDE Frameworks Version: 5.79.0 Qt Version: 5.15.2
FYI you can adjust the opacity of the fill color of the lines, for example the plasma widgets use more transparency by default
Please don't put multiple issues in one report, it makes it hard to deal with the individual issues. > But KSysguard solved that more nicely, by first shrinking the legend to just colored percentages. You should still wrap them at some point, because otherwise 48 threads will fit on pretty much no screen, but having a colored line next to it is not very useful at that size. This may be tweaked later, but KSysGuard's solution also does not scale too well and additionally has quite some contrast problems. > Secondly, the default of using a stacked chart leads to a chart going to 1600%, but because only one core is pinned, that one sits at 1/16th of the chart. This does not provide much useful information about why an application hangs, since it looks like it is hanging in IO instead of in a CPU loop. Seems to me this is better solved by the process table? How would you even distinguish single process usage from the history chart? In any case, stacked means you get to see individual core usage as well as overall usage and in my opinion it is actually easier to distinguish cores, even with lower fill opacity the screenshot of KSysGuard is pretty useless to me. That said, this is why the entire thing is configurable so you can choose not to use stacked (and as David said, you can also choose the fill opacity.)
*** Bug 442509 has been marked as a duplicate of this bug. ***
*** Bug 463761 has been marked as a duplicate of this bug. ***
Re-opening as we continue to accumulate complaints about this. It's notable that System Monitor has opposite default settings compared to ksysguard did: the CPU percentage in the list view is averaged rather than separate, while the CPU graph view is combined rather than averaged. It seems that people preferred the prior ksysguard default settings, not the new ones for System Monitor.
A possibly relevant merge request was started @ https://invent.kde.org/plasma/plasma-systemmonitor/-/merge_requests/206
*** Bug 490895 has been marked as a duplicate of this bug. ***
*** Bug 501152 has been marked as a duplicate of this bug. ***
Adding to the dogpile of complaints. The way it worked in Ksysguard was better than System Monitor in Plasma 6.3.4.
In Plasma 6.4, there's now an averaged CPU usage graph on the History page!
*** Bug 512366 has been marked as a duplicate of this bug. ***