x86/vf_blend: Add SSE2 optimization for divide
authorTimothy Gu <timothygu99@gmail.com>
Wed, 10 Feb 2016 09:04:51 +0000 (09:04 +0000)
committerTimothy Gu <timothygu99@gmail.com>
Sun, 28 Feb 2016 16:19:09 +0000 (08:19 -0800)
commit222e6da605eadd9afa386f0a6c3142b16e16cf74
treee701f0c4485279d065245ae87571a7109a224177
parent1c9215e580b6436d1aff3c0118ef01269712ebd9
x86/vf_blend: Add SSE2 optimization for divide

 4.5x faster than C float version with autovectorization
10  x faster than C int version
25  x faster than C float version without autovectorization
libavfilter/x86/vf_blend.asm
libavfilter/x86/vf_blend_init.c