Use more MAC16_16's and unroll a loop.
authorTimothy B. Terriberry <tterribe@xiph.org>
Wed, 22 May 2013 22:26:12 +0000 (15:26 -0700)
committerTimothy B. Terriberry <tterribe@xiph.org>
Wed, 22 May 2013 22:32:50 +0000 (15:32 -0700)
commit85ede2c6aa066da29fce5186394f46927358be3b
tree3aeddc60d6a37142b2faa66f1041a02dc91bf505
parent2040606f4a1d3b230bdf00e1b7e4427df8dcdd3b
Use more MAC16_16's and unroll a loop.

This splits out the non-arch-specific portions of a patch written
 by AurĂ©lien Zanelli <aurelien.zanelli@parrot.com
 http://lists.xiph.org/pipermail/opus/2013-May/002088.html

I also added support for odd n, for custom modes.

0.25% speedup on 96 kbps stereo encode+decode on a Cortex A8.
celt/celt_lpc.c