[MPlayer-dev-eng] Re: Make -nortc the default?

[Diego Biurrun]

On Sat, Sep 09, 2006 at 12:07:08PM -0400, Rich Felker wrote:
> On Fri, Sep 08, 2006 at 06:12:46PM +0000, Carl Eugen Hoyos wrote:
> > Uoti Urpala <uoti.urpala <at> pp1.inet.fi> writes:
> > 
> > > Does anyone have a reason why rtc timing should still be used by
> > > default?
> > > 
> > > Reasons not to use it:
> > > - Most desktop Linux boxes have HZ 1000 or at least 250 nowadays. HZ
> > >   1000 gives accuracy comparable to RTC using normal sleep, and 250 is
> > >   probably still high enough. Thus rtc doesn't give any benefit in most
> > >   cases.
> > > - rtc wakes up the process 1024 times a second while sleeping, wasting
> > >   CPU that could be used for other processes
> > > - rtc requires the user to set special permissions for /dev/rtc/, and if
> > >   that hasn't been done it shows error messages telling the user to
> > >   modify system startup scripts.
> > 
> > - -nortc typically saves 10% CPU time on playback.
> >   That often makes the difference when trying to play HD content (or just H264).
> >   (That's on an AMD 2600+ running SuSE 10.0, 2.6.13)
> 
> Seems to be the same on my box. I haven't tested whether -nortc makes
> some files that were previously unplayable playable, however.

I seem to get similar results on my K6-III with some unscientific
benchmark.  Playing a H.264 sample that is a bit too complex for my box
drops slightly less frames with -nortc than with -rtc.

Diego