1 ;*****************************************************************************
2 ;* MMX/SSE2/AVX-optimized 10-bit H.264 chroma MC code
3 ;*****************************************************************************
4 ;* Copyright (C) 2005-2011 x264 project
6 ;* Authors: Daniel Kang <daniel.d.kang@gmail.com>
8 ;* This file is part of Libav.
10 ;* Libav is free software; you can redistribute it and/or
11 ;* modify it under the terms of the GNU Lesser General Public
12 ;* License as published by the Free Software Foundation; either
13 ;* version 2.1 of the License, or (at your option) any later version.
15 ;* Libav is distributed in the hope that it will be useful,
16 ;* but WITHOUT ANY WARRANTY; without even the implied warranty of
17 ;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 ;* Lesser General Public License for more details.
20 ;* You should have received a copy of the GNU Lesser General Public
21 ;* License along with Libav; if not, write to the Free Software
22 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 ;******************************************************************************
25 %include "libavutil/x86/x86util.asm"
37 %macro MV0_PIXELS_MC8 0
43 CHROMAMC_AVG m0, [r0 ]
44 CHROMAMC_AVG m1, [r0+r2 ]
49 CHROMAMC_AVG m0, [r0+r2*2]
50 CHROMAMC_AVG m1, [r0+r4 ]
59 ;-----------------------------------------------------------------------------
60 ; void ff_put/avg_h264_chroma_mc8(pixel *dst, pixel *src, int stride, int h,
62 ;-----------------------------------------------------------------------------
64 cglobal %1_h264_chroma_mc8_10, 6,7,8
68 jne .at_least_one_non_zero
69 ; mx == 0 AND my == 0 - no filter needed
73 .at_least_one_non_zero:
77 mov r6, r2 ; dxy = x ? 1 : stride
81 ; mx == 0 XOR my == 0 - 1 dimensional filter only
85 mova m6, [pw_4] ; mm6 = rnd >> 3
86 SPLATW m5, m5 ; mm5 = B = x
87 psubw m4, m5 ; mm4 = A = 8-x
90 movu m0, [r1 ] ; mm0 = src[0..7]
91 movu m2, [r1+r6] ; mm2 = src[1..8]
93 pmullw m0, m4 ; mm0 = A * src[0..7]
94 pmullw m2, m5 ; mm2 = B * src[1..8]
100 mova [r0], m0 ; dst[0..7] = (A * src[0..7] + B * src[1..8] + (rnd >> 3)) >> 3
108 .xy_interpolation: ; general case, bilinear
112 SPLATW m4, m4 ; mm4 = x words
113 SPLATW m6, m6 ; mm6 = y words
114 psllw m5, m4, 3 ; mm5 = 8x
115 pmullw m4, m6 ; mm4 = x * y
116 psllw m6, 3 ; mm6 = 8y
117 paddw m1, m5, m6 ; mm7 = 8x+8y
118 mova m7, m4 ; DD = x * y
119 psubw m5, m4 ; mm5 = B = 8x - xy
120 psubw m6, m4 ; mm6 = C = 8y - xy
122 psubw m4, m1 ; mm4 = A = xy - (8x+8y) + 64
124 movu m0, [r1 ] ; mm0 = src[0..7]
125 movu m1, [r1+2] ; mm1 = src[1..8]
131 paddw m2, m1 ; mm2 = A * src[0..7] + B * src[1..8]
136 paddw m2, m3 ; mm2 += C * src[0..7+strde]
138 paddw m2, m3 ; mm2 += D * src[1..8+strde]
142 CHROMAMC_AVG m2, [r0]
143 mova [r0], m2 ; dst[0..7] = (mm2 + 32) >> 6
151 ;-----------------------------------------------------------------------------
152 ; void ff_put/avg_h264_chroma_mc4(pixel *dst, pixel *src, int stride, int h,
154 ;-----------------------------------------------------------------------------
170 CHROMAMC_AVG m1, %2, [r0]
176 cglobal %1_h264_chroma_mc4_10, 6,6,7
202 ;-----------------------------------------------------------------------------
203 ; void ff_put/avg_h264_chroma_mc2(pixel *dst, pixel *src, int stride, int h,
205 ;-----------------------------------------------------------------------------
207 cglobal %1_h264_chroma_mc2_10, 6,7
213 imul r5d, r4d ; x*y<<16 | y*(8-x)
215 sub r4d, r5d ; x*(8-y)<<16 | (8-x)*(8-y)
219 punpckldq m5, m5 ; mm5 = {A,B,A,B}
220 punpckldq m6, m6 ; mm6 = {C,D,C,D}
222 pshufw m2, [r1], 0x94 ; mm0 = src[0,1,1,2]
227 pmaddwd m1, m5 ; mm1 = A * src[0,1] + B * src[1,2]
228 pshufw m0, [r1], 0x94 ; mm0 = src[0,1,1,2]
232 paddw m1, m0 ; mm1 += C * src[0,1] + D * src[1,2]
235 CHROMAMC_AVG m1, m3, [r0]
252 %define CHROMAMC_AVG NOTHING
261 %define CHROMAMC_AVG AVG