On 11/1/05, Michaelwagner
<[EMAIL PROTECTED]> wrote:
>
> I think what's needed here is a bit more management of expectations,
> that's all. SLiMP3 support in 6.2 works for most people. It doesn't
> seem to work for this one person.

More than one person.

As I mentioned, I know multiple people who have the original slimp3
and who cannot go to 6.X for performance and bug reasons.   They've
tried multiple times and GIVEN UP because they're always disappointed
and have to go back.

We were told the database would be a HUGE improvement and waited years for it..

As to the rescan taking a long time..  Yep, I waited until it was
done.  My slimp3 is hardwired.  My library is 40K songs. I do a lot of
transcoding from flac (flac>wav>mp3).  Dedicated XP2200/512 MB, Gentoo
linux.

As for 'quit bitching and debug', I've posted extensive performance
data here in the past and sent it to support .  Not much response.
Some improvements were made.  HOWEVER, the core problem and symptoms
remain unsolved.  I don't think the test harness does much to measure
and track performance. Otherwise, these perf issues wouldn't be
getting out.  Have the specifics of the 'performance test harness and
process' ever been disclosed?

This is a streaming media delivery architecture which is implemented
as a MONOLITHIC process.  The nearly constant addition of features to
the monolithic arch presents an on-going threat to erode performance.

A lot of the performance problems would go away if there was a
separate process which did nothing but deliver music to the slimp3/sb
(and handle display features, etc). I think this should be broken out
from the main server.  Another approach, adding threads to the main
server, would probably have too negative an impact due to bugs as a
result of increased code complexity, etc.  Once completed, the content
delivery code should be rock solid and frozen.

Until the monolithic issue is resolved, it seems like the work on
performance is just trying to bandaid symptoms.  So that is where a
lot of the frustration is coming from.. We've been waiting on core
arch improvements for YEARS and they just haven't come.  I'm not aware
if there is even a plan to resolve these low level issues. 
Performance will be an on-going problem until the arch issues are
resolved.  Increasing the client buffer size is one way to treat the
symptoms of the problems but it doesn't work for older devices and it
is another bandaid.


Regards,

  FL
_______________________________________________
Discuss mailing list
[email protected]
http://lists.slimdevices.com/lists/listinfo/discuss

Reply via email to