Music Player Review – Power Usage ~ wakeups per second

For the first performance head to head, we take the remaining music players and test their ability to not wakeup the processor unnecessarily.  More wakeups per second is bad and means that the processor can’t go to sleep, this usually translates to more power usage. Powertop was used for the testing.

The first graph (below) shows the startup, playing, and idle after playing wakeups per second on a logarithmic scale.    The “playing” wake ups don’t really matter.

This second graph goes more into detail on the two key data points, startup and idle wakeups.

The Best:   Rhythmbox and Musique get practically perfect scores

Second Best: Clementine and Amarok were even at about 1.1 wakeups per second (which is still really quite good)

The Middle:  Gmusicbrowser did the best, followed closely by Quod Libet and then Banshee.  These are still relatively good results from powertop.

The Worst:  Guayadeque clearly lost where it counts, but Audacious didn’t do so well either.

Testing notes: Banshee froze (<30 seconds) on startup updating something.  Amarok spent a good amount of time at startup writing out to files (powertop picked up on this).

I’m not removing any music players in this round.  I’m going to wait until the memory tests (next), and the CPU tests are done before choosing which players don’t get to make it to the next round of review.

11 thoughts on “Music Player Review – Power Usage ~ wakeups per second

    1. VLC is well behaved: startup/idle .1.

      You can find results yourself by doing something like: sudo powertop-1.13 -d 10 | grep vlc

  1. 1. What versions of the applications are you using?

    2. Why do you consider wakeups while playing irrelevant but consider wakeups during startup important? Seems like a player should spend a lot more time playing music than starting up; hence wakeups while playing is much more relevant to overall power usage.

    P.S. My results with Audacious are vastly different than yours — between 1.1 and 1.5 wakeups/second at idle with Audacious 3.3.1+git.

    1. 1. All the stock versions from Ubuntu 12.04. Audacious is 3.2.1
      2. My fault, by startup I actually mean after startup, not during (Amarok would be the worst by far in that category, btw). I’m expecting CPU usage to be a better indicator when it’s actually playing. All of them need to wake the processor up to play music. The idle time is when you pause the music does the music player let the computer processor “rest”. Recording both just let’s us see if the player is consistent.

      I used the old version of powertop-1.13 (the beta one in Ubuntu seemed buggy), specifically (I think the actual results were from “-d 30″):
      Startup# sudo powertop-1.13 -d 10 | grep audacious
      0.2% ( 4.0) audacious
      Playing# sudo powertop-1.13 -d 10 | grep audacious
      3.2% ( 86.2) audacious
      Idle# sudo powertop-1.13 -d 10 | grep audacious
      0.2% ( 4.0) audacious

      If this is no longer the case in git, that’s great :)

  2. Hmm, I guess I have a buggy version of powertop (2.1); tried 1.13 and I get the same 4 wakeups/second as you — but your charts show 8, not 4.

    I am always interested in making things better. The 4 wakeups/second is the timeout for monitoring the system volume — do you think that should be reduced? I tend to consider anything less than 10 as insignificant. (Firefox sitting idle with only your blog open generates about 90 …)

    1. Oddly enough the second time I ran it showed 8. I am using the library functionality if that could be significant.

      8 or 4 wakeups by themselves certainly won’t be significant, however if every app says that then all of a sudden they are :). And yes Firefox has been very bad at wakeups, but it seems like it might be improving (my nightly version shows <1 wakeup per second).

      Why do you need to check the system volume at all? And why 4 times a second?

      1. To keep the volume slider in the UI up to date. Anything less than 4 times a second seems laggy/jumpy to me.

        Just checked again and I’m now also seeing 8 wakeups/second when a song is paused. Guess I have something to look into.

Add Comment Register

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>