#5988 HIGH FutureF: Theora encoding is too slow

[Zarro Boogs per Child]

#5988: Theora encoding is too slow
--------------------------+-------------------------------------------------
  Reporter:  bemasc       |       Owner:  jg            
      Type:  enhancement  |      Status:  new           
  Priority:  high         |   Milestone:  FutureFeatures
 Component:  distro       |     Version:                
Resolution:               |    Keywords:                
  Verified:  0            |    Blocking:                
 Blockedby:               |  
--------------------------+-------------------------------------------------

Comment(by bemasc):

 {{{
 (20:59:55) bemasc: xiphmont: If I'm trying to make Theora go fast on a
 OLPC XO, should I be testing thusnelda?
 (21:12:31) xiphmont: bemasc: Yes, but in the mid-term.  Thusnelda is not
 yet ready for any real lifting, and it is also in a dishevelled state
 (21:12:57) bemasc: ok.  Testing, but only for future-planning purposes.
 Cool.
 (21:12:58) xiphmont: Meaning, don't go nuts with assebly optimization or
 the like--- as it would all get blown up again in the near term.
 (21:13:23) bemasc: oh.  So testing, with the idea that I might be able to
 create a stable build of thusnelda, with some care.
 (21:13:45) xiphmont: it's not so much that it's unstable as there's alot
 of fundamental work left to do.
 (21:13:53) xiphmont: I can tell you what is likely safe and what isn't.
 (21:14:09) bemasc: That would be good to know, I suppose.
 (21:15:17) bemasc: But if I just attempt a usual sort of build with
 sensible CFLAGS, you would expect it to work, and be a bit faster than
 standard Theora?
 (21:16:50) xiphmont: oh
 (21:16:52) xiphmont: yes
 (21:17:13) bemasc: lovely
 (21:17:27) xiphmont: eventually it will be much faster.  Right now,
 depending on what you code, it is about the same or alot better.
 (21:17:57) xiphmont: However, current Thusnelda HEAD has a token packing
 bug that pops up now and then.  Be careful using it to code anything you
 want to keep ;-)
 (21:18:09) bemasc: hmm
 (21:18:22) bemasc: that's not a good kind of bug
 (21:18:42) xiphmont: No.  I have to bisection search/kill it first thing
 tomorrow.
 (21:18:53) xiphmont: I noticed it about half an hour before last week's
 big demo.
 (21:19:15) xiphmont: Had no chance yet to look at it.  I expect it will
 not be hard to find.  It's bleeding-recent.
 }}}

 Sounds like we should maybe also be testing the Thusnelda branch.

-- 
Ticket URL: <http://dev.laptop.org/ticket/5988#comment:4>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system