From ca5c3ff90972a5c97aabda2ace57ba72dcd7d83b Mon Sep 17 00:00:00 2001 From: Michael Niedermayer Date: Sat, 15 Nov 2014 04:07:08 +0100 Subject: vf_interlace: x86: improve asm performance 4775 decicycles -> 3688 decicycles --- libavfilter/x86/vf_interlace.asm | 46 +++++++++++++++++----------------------- 1 file changed, 19 insertions(+), 27 deletions(-) (limited to 'libavfilter') diff --git a/libavfilter/x86/vf_interlace.asm b/libavfilter/x86/vf_interlace.asm index 8c2e9b099c..b8d86165bf 100644 --- a/libavfilter/x86/vf_interlace.asm +++ b/libavfilter/x86/vf_interlace.asm @@ -2,6 +2,7 @@ ;* x86-optimized functions for interlace filter ;* ;* Copyright (C) 2014 Kieran Kunhya +;* Copyright (c) 2014 Michael Niedermayer ;* ;* This file is part of Libav. ;* @@ -34,36 +35,27 @@ cglobal lowpass_line, 5, 5, 7 add r4, r1 neg r1 - pxor m6, m6 + pcmpeqb m6, m6 .loop - mova m0, [r2+r1] - punpcklbw m1, m0, m6 - punpckhbw m0, m6 - paddw m0, m0 - paddw m1, m1 + mova m0, [r3+r1] + mova m1, [r3+r1+mmsize] + pavgb m0, [r4+r1] + pavgb m1, [r4+r1+mmsize] + mova m2, [r2+r1] + mova m3, [r2+r1+mmsize] + pxor m0, m6 + pxor m1, m6 + pxor m2, m6, [r2+r1] + pxor m3, m6, [r2+r1+mmsize] + pavgb m0, m2 + pavgb m1, m3 + pxor m0, m6 + pxor m1, m6 + mova [r0+r1], m0 + mova [r0+r1+mmsize], m1 - mova m2, [r3+r1] - punpcklbw m3, m2, m6 - punpckhbw m2, m6 - - mova m4, [r4+r1] - punpcklbw m5, m4, m6 - punpckhbw m4, m6 - - paddw m1, m3 - pavgw m1, m5 - - paddw m0, m2 - pavgw m0, m4 - - psrlw m0, 1 - psrlw m1, 1 - - packuswb m1, m0 - mova [r0+r1], m1 - - add r1, mmsize + add r1, 2*mmsize jl .loop REP_RET %endmacro -- cgit v1.2.3