Bug 238331 - CPU utilization includes IO-wait
Summary: CPU utilization includes IO-wait
Status: RESOLVED INTENTIONAL
Alias: None
Product: ksysguard
Classification: Applications
Component: general (show other bugs)
Version: unspecified
Platform: Ubuntu Linux
: NOR normal
Target Milestone: ---
Assignee: John Tapsell
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-05-20 20:36 UTC by Søren Holm
Modified: 2023-02-22 18:17 UTC (History)
3 users (show)

See Also:
Latest Commit:
Version Fixed In:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Søren Holm 2010-05-20 20:36:36 UTC
Version:            (using KDE 4.4.3)
OS:                Linux
Installed from:    Ubuntu Packages

The cpu-monitor shows cpu-utillization + IO-wait time. This is wrong, since IO-wait is not actually used CPU-cycles.
Comment 1 Aaron J. Seigo 2010-05-21 04:47:41 UTC
it uses ksyguard for this information.
Comment 2 Søren Holm 2010-08-14 00:16:14 UTC
Can the display be changed so it only includes actual used cpu-cycles?
Comment 3 Søren Holm 2012-09-14 17:34:05 UTC
Could you please fix the bug or point be in the direction of where the calculation is done. It is realy annoying when duing io from slow medias or over sshfs-mounts. The actual indication then maxes out one of my cpu-cores.
Comment 4 Andrew Crouthamel 2018-09-05 12:53:07 UTC
Hi there. CPU usage in Linux includes iowait. This is because iowait is time that the processor is waiting during which there was outstanding disk I/O requests for that processor. So that is the cycles of the CPU unusable for calculations, and thus considered "used".
Comment 5 Russell Haley 2023-02-15 14:08:45 UTC
(In reply to Andrew Crouthamel from comment #4)
>So that is the cycles of the CPU unusable for calculations, and thus considered "used".

I/O wait does not make the CPU unusuable for other calculations.  It's just idle time that happens when some I/O request is in flight. Try running my reproducer script from https://bugs.kde.org/show_bug.cgi?id=465768 and "stress -c $(nproc)" at the same time.  Top will show 100% userspace utilization from the stress command, *and* the disk reads will continue at full speed.