use xorps instead of mulps to toggle the sign of a float, as suggested by Software...
authorZuxy Meng <zuxy.meng@gmail.com>
Sun, 5 Mar 2006 20:25:18 +0000 (20:25 +0000)
committerGuillaume Poirier <gpoirier@mplayerhq.hu>
Sun, 5 Mar 2006 20:25:18 +0000 (20:25 +0000)
commit2ffb22d2ad014e7fa8ee4c24263eb03abb9e86e5
tree0108b52940ff24d5e18fbd4b56f5f1fc52cb2a61
parentd84f7c61eede0e5327c24e606097c271abc7fea3
use xorps instead of mulps to toggle the sign of a float, as suggested by Software Optimization Guide for AMD64 Processors.
Patch by Zuxy Meng < zuxy POIS meng AH  gmail POIS com > OKed by Michael
Original thread:
Date: Mar 5, 2006 8:15 PM
Subject: [Ffmpeg-devel] [PATCH] Little optimization to fft_sse.c

Originally committed as revision 5112 to svn://svn.ffmpeg.org/ffmpeg/trunk
libavcodec/i386/fft_sse.c