lavr: replace the SSE version of ff_conv_fltp_to_flt_6ch() with SSE4 and AVX
authorJustin Ruggles <justin.ruggles@gmail.com>
Thu, 3 May 2012 19:23:32 +0000 (15:23 -0400)
committerJustin Ruggles <justin.ruggles@gmail.com>
Wed, 9 May 2012 20:17:59 +0000 (16:17 -0400)
commit5cc6d5244d4ec89b3ac855abff4a3d19caee22f1
tree7c3643f535a67203b60bd001f1382b9b1282d3af
parent0b45334a5880d6e2a4b3642adcd5feab8a27a150
lavr: replace the SSE version of ff_conv_fltp_to_flt_6ch() with SSE4 and AVX

The current SSE version is slower than the MMX version on Athlon64 and Sandy
Bridge, but the SSE4 and AVX versions are faster on Sandy Bridge.
libavresample/x86/audio_convert.asm
libavresample/x86/audio_convert_init.c
libavutil/x86/x86util.asm