[vorbis-dev] vorbis on playstation 2
etherton at rockstarsandiego.com
Thu Feb 6 19:52:43 PST 2003
So I ran some tests with the saner version of MULT31 and it knocked the
original cycle count of 950M down to 900M, more than the benefit that moving
the LUT into scratchpad gave me.
Reintroducing the scratchpad code dropped the total cycle count to 853M.
The PS2 has dual-integer multipliers, but requires a specific "pipe one"
insn that the compiler itself won't generate that I'll need to look into.
It looks like the window function _vorbis_apply_window should be pretty easy
to speed up 2x as well.
I guess I need to take a closer look at the vorbis spec, but I'm not clear
when the window is applied. Is there a way I could change the data used by
vorbis_apply_window to 16-bits to save on D$ misses and make it able to use
the eight-wide MMI? The MMI also has saturating arithmetic, which might
help out the final PCM conversion as well.
Thanks again for everybody who's been helping me out here!
--- >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.
More information about the Vorbis-dev