Red Hat Bugzilla – Bug 90760
Screensavers consuming too much CPU
Last modified: 2014-03-16 22:36:11 EDT
I did see a few issues still, but none of them are as serious as before. A couple still chew up a lot of cpu. -"Maze" does not work, and causes X to chew up to 99.9% of the cpu, but the difference from before is that the machine is still usable, before it wasn't. - Decayscreen chews up 80-89% cpu. Not sure if you care about these other screen saver issues, but since I saw them I thought I'd mention them even though there is no real impact. - Crystal will cause any screen savers run after it to not work - Flag is in the list twice - GLPlanet is blank on GX260s - Lisa & Noseguy don't work, just show previous screen saver not moving - Pipes is just flickers on GX260s I only tried on Gnome since that is where I saw the issues before. ---------- Action by: mwaite Issue Registered ---------- Action by: mwaite Searching Bugzilla now for previous entries. mwaite assigned to issue for Nortel/CSC. Category set to: X Windows::Gnome Status set to: Waiting on Tech
Many of the screensavers try to draw as fast as possible... this will cause them to use CPU. If this is not desired, pick a different one, or, for instance, a blank screen. The GX260-related problems sound like graphics card specific issues.
For the record, *none* of the screensavers try to draw as fast as possible: all of them (by default) spend enough time in usleep() that they should not be loading the system significantly. Plus, xscreensaver runs all of them at a low priority with nice. (The unfortunate exception to this is the case of OpenGL hacks running on systems without GL hardware, in which case, the X server itself tremendously loads the machine.) However, it would be normal for a screensaver to consume 80%+ of the CPU -- if the CPU would otherwise have been *completely idle*. That is, the saver should never make the load go above 1.0, though *approaching* 1.0 would be expected.
Closing as not a bug; as jwz said things are acting more or less as designed.