Re: [xine-user] upgrades

[Dave P. <dl...@br...>]

Hi Bill:

  First of all, thanks a lot for helping me with xine. The 'xine -V xv'
seems to make a considerable difference, dropping CPU usage down to a
much better 20-30%. What does xine utilize for its default video output
driver ? The upgrades may have helped too. DMA is indeed enabled, and
I've included instructions for doing so in my article.

  Interesting note re: your disk caching experience. I haven't tried to
replicate it but will do so and report back on my experience.

  Btw, does xine support chapter selection ? I must be missing it if
it's in there somewhere... :)

Best regards,

== Dave Phillips

	The Book Of Linux Music & Sound at http://www.nostarch.com/lms.htm
	The Linux Soundapps Site at http://linux-sound.org

Currently listening to: "O ignis spiritus Paracliti" (Hildegard von
Bingen)


Bill Fink wrote:
> 
> Hi Dave,
> 
> On Sat, 24 May 2003, Dave Phillips wrote:
> 
> > Btw, a curious note: xine scored highest in my CPU-usage comparison
> > during the tests, is there some way to significantly reduce its need for
> > processor cycles ? FYI, the scores from lowest to highest went :
> >
> >       VideoLAN Client 20-30%
> >       Ogle    20-40%
> >       MPlayer 40-50%
> >       xine    50-70%
> >
> > These are approximate average low-to-high percentages reported by
> > gkrellm during playback for the Blade Runner DVD. Same load on the
> > machine for all tests, but I emphasize that this is hardly scientific
> > reporting on my part and I welcome any suggestions for performance
> > improvement.
> 
> The two biggest performance hits I know of are not having DMA enabled
> and using Xshm instead of Xv.  Not having DMA enabled should affect all
> the applications the same.  It is possible that xine is using Xshm
> instead of Xv.  Try running xine once as "xine -V xv" to force use
> of Xv (this is a sticky setting and xine will remember it for future
> runs).  Also try running xine-check to see if it gives any hints about
> performance settings.
> 
> I've also noticed a weird disk caching bug/behavior that xine (and
> perhaps others) exposes.  On my system, if I run xine immediately after
> rebooting and logging into kde, xine runs less well than usual, with
> noticeable jerkiness in DVD video playback and consuming considerably
> more CPU than normal (about 60% more).  If I just copy a large 102 MB
> file to another file and then rerun xine, xine then runs well and
> consumes much less CPU.  This behavior is very repeatable.  A friend
> theorized that the file copy caused the amount of memory to be used
> for disk caching to be substantially increased, and thus the kernel
> didn't need to do any memory allocations for the disk cache when xine
> was subsequently used to play the DVD.  I thought there might be a disk
> caching bug affecting X somehow, since X is the last thing started
> during the boot process, and that copying the large file cleared any
> bogus information out of the disk cache, and that's why xine then
> worked normally afterward.  In any event, this could cause the order
> of running your tests to be significant.  Did you happen to run xine
> either first or shortly after booting?
> 
>                                                 -Bill