| Summary: | System Monitor reports misleadingly low cpu usage | ||
|---|---|---|---|
| Product: | [Applications] plasma-systemmonitor | Reporter: | John van Spaandonk <jwork123nl> |
| Component: | general | Assignee: | KSysGuard Developers <ksysguard-bugs> |
| Status: | ASSIGNED --- | ||
| Severity: | normal | CC: | ahiemstra, aspotashev, ddascalescu+kde, nate, plasma-bugs-null, postix |
| Priority: | NOR | Keywords: | usability |
| Version First Reported In: | 5.23.90 | ||
| Target Milestone: | --- | ||
| Platform: | Neon | ||
| OS: | Linux | ||
| See Also: | https://bugs.kde.org/show_bug.cgi?id=434040 | ||
| Latest Commit: | Version Fixed/Implemented In: | ||
| Sentry Crash Report: | |||
|
Description
John van Spaandonk
2022-01-31 14:51:57 UTC
Yes, per-core usage is usually divided by core count. You can change it by right clicking on the header and uncheck "divide by number of cpus". What the correct default is is very debatable. Thanks for the advice, may I say that this right-mouse button click is not very discoverable? I see now how I need to interpret the tooltip "The current total CPU usage of a process" as "Process CPU usage relative to available CPU". Although you can reason that this is not true since if a program is designed to work only on 1 CPU, the available CPU is only 1 CPU. In other words: I guarantee that a user will not understand why his softsynth is not working because it hugs one CPU while it is being reported as only 25%. So then let me ask this: why do you choose a different convention than top? Why not be in line with what is used on the command line? *** Bug 468106 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 separate rather than averaged. It seems that people preferred the prior ksusguard 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 I still see this behavior. I've recorded a video with my phone during a Google Meet call running in Firefox. It was impossible to identify which process was consuming the CPU. https://youtu.be/8ogXjP1kdRg Operating System: Fedora Linux 38 KDE Plasma Version: 5.27.10 KDE Frameworks Version: 5.113.0 Qt Version: 5.15.11 Kernel Version: 6.7.7-100.fc38.x86_64 (64-bit) Graphics Platform: Wayland Processors: 20 × 12th Gen Intel® Core™ i7-12700H Memory: 62.5 GiB of RAM Graphics Processor: Mesa Intel® Graphics Manufacturer: TUXEDO Product Name: TUXEDO InfinityBook Pro Gen7 (MK1) Yeah, the maintainer didn't like my patch and I haven't succeeded in convincing him yet. To be fair, I haven't tried again in a while. *** Bug 504994 has been marked as a duplicate of this bug. *** This problem gets worse over time as core counts increase. For example, on my consumer-grade 16-core laptop system that I bought for under $600, a process that displays 6% CPU usage is actually maxing out an entire core and generating a substantial amount of heat and fan noise, but this rogue process won't be easily identifiable based on its CPU usage number. I continue to think that scaling the per-process numbers to 100% is not a useful default state. A possibly relevant merge request was started @ https://invent.kde.org/plasma/plasma-systemmonitor/-/merge_requests/372 |