[vorbis-dev] mdct_backward with fused muladd?

[David Etherton]

> No.  See the definition of MULT32() and MULT31() for the _LOW_ACCURACY_
> case.  I had to make the multiplication unbalanced (like 24x8) since a
16x16
> would not give acceptable audio quality.

This reminds me -- the cosine lut looks like it's a 0.31 int array (but all
values are unsigned).  What's the expected range of the pcm input data to
mdct_backward, and where is the binary fixed point?  Can anybody recommend
any good books or web links that cover audio compression theory (I know a
little bit about dct's from jpeg compression but I'm not sure about audio)
so I can stop asking so many newbie questions?

Also, if anybody cares I'd be glad to share the work I've done on the PS2 so
far (as I discussed in several older emails, it's mostly from using the
dual-pipe integer multiplies in the XPROD macros; the lion's share of the
cpu load is in mdct_backward).

-Dave

--- >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.