[vorbis-dev] vorbis on playstation 2

[Segher Boessenkool]

David Etherton wrote:
>>Depending on the stream used, the second table is unlikely to be
> 
> referenced.
> 
> That reminds me -- looking at the profile hits in mdct_backward, there were
> several switch statement cases that never got any hits.  Given that I don't
> need to decode any arbitrary ogg vorbis bitstream, are there simplications I
> can make here to cut down on the amount of code I have to optimize?

Throw away floor0 and res0 and res1.  Only keep floor1
and res2.  Special case all the transform stuff to the
two blocksizes used (if you use two of-em at all).

You can also make smaller or algoritmically decodable
Huffman tables instead of the big lookups, if you so
choose (you'll have to tune the encoder to generate
correct streams for you, in that case, though).  You
can even hardwire the modes into the decoder (no headers
needed anymore --> good for small samples, no setup
overhead).

This is open source, anything is possible :)

<p>Segher

<p>--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.