[Theora-dev] Re: Questions, MMX and co.

[Adam D. Moss]

Ralph Giles wrote:
>>5) How to *validate* the implementation? It is probably easy to introduce 
>>biais in the encoder in the C to MMX transformation process. On the other 
>>hand, sometimes output is different and this is normal (psavgb does (A+B+1)/2 
>>which is more precise than (A+B)/2). Maybe we can simply trust experimental 
>>work, or rely on a good 4), but then...?
> 
> Well, encoder output should be bit-for-bit identical, at least for the 
> reference implementation. On the decoder one can make speed/quality 
> tradeoffs.

Is this the right way around?  I thought the DEcoder had to be
bit-for-bit identical because that's what VP3 assumes (precise
decoding rules allow unbound GOP size, because of supposedly no
cumulative error).

The MMXized ENcoder as it stands is certainly far from bit-for-bit
identical with the output of the reference encoder -- even its own
output between runs is non-deterministic (which probably isn't a good sign).

--Adam