[theora-dev] Multi-Thread Theora Encoder

Felipe Portavales Goldstein portavales at gmail.com
Wed Oct 10 08:46:38 PDT 2007

On 10/7/07, Maik Merten <maikmerten at gmx.net> wrote:
> Oh, another thing I noticed:
> encode.c, PickModes() (starting with line 1439) has a few mallocs for
> your multithreading data structures. If the return statement in line
> 1467 gets triggered the memory allocated doesn't reveive the free()
> calls it deserves. (if I don't miss something, one can rightfully call
> me a C newbie).

Thanks for noting that. This is a really serious bug.

> The return statement in line 1467 is a questionable hack to skip motion
> search for the fastest speed level. Thinking about it I think it should
> perhaps set the error scores to something different to zero (I assume
> zero means "perfect match found" while when skipping motion compensation
> the result should be "no match found").

This is an issue relative to the original encoder also.
My patch does not introduce this hack.
But you're right, it is something that should be discussed.

Thanks Maik


> Maik
> _______________________________________________
> theora-dev mailing list
> theora-dev at xiph.org
> http://lists.xiph.org/mailman/listinfo/theora-dev

Felipe Portavales Goldstein <portavales at gmail>
Undergraduate Student - IC-UNICAMP
Computer Systems Laboratory

More information about the theora-dev mailing list