2 * RV40 decoder motion compensation functions x86-optimised
3 * Copyright (c) 2008 Konstantin Shishkov
5 * This file is part of Libav.
7 * Libav is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * Libav is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
24 * RV40 decoder motion compensation functions x86-optimised
27 #include "libavcodec/rv34dsp.h"
29 void ff_put_rv40_chroma_mc8_mmx (uint8_t *dst, uint8_t *src,
30 int stride, int h, int x, int y);
31 void ff_avg_rv40_chroma_mc8_mmx2 (uint8_t *dst, uint8_t *src,
32 int stride, int h, int x, int y);
33 void ff_avg_rv40_chroma_mc8_3dnow(uint8_t *dst, uint8_t *src,
34 int stride, int h, int x, int y);
36 void ff_put_rv40_chroma_mc4_mmx (uint8_t *dst, uint8_t *src,
37 int stride, int h, int x, int y);
38 void ff_avg_rv40_chroma_mc4_mmx2 (uint8_t *dst, uint8_t *src,
39 int stride, int h, int x, int y);
40 void ff_avg_rv40_chroma_mc4_3dnow(uint8_t *dst, uint8_t *src,
41 int stride, int h, int x, int y);
43 #define DECLARE_WEIGHT(opt) \
44 void ff_rv40_weight_func_16_##opt(uint8_t *dst, uint8_t *src1, uint8_t *src2, \
45 int w1, int w2, ptrdiff_t stride); \
46 void ff_rv40_weight_func_8_##opt (uint8_t *dst, uint8_t *src1, uint8_t *src2, \
47 int w1, int w2, ptrdiff_t stride);
52 void ff_rv40dsp_init_x86(RV34DSPContext *c, DSPContext *dsp)
55 int mm_flags = av_get_cpu_flags();
57 if (mm_flags & AV_CPU_FLAG_MMX) {
58 c->put_chroma_pixels_tab[0] = ff_put_rv40_chroma_mc8_mmx;
59 c->put_chroma_pixels_tab[1] = ff_put_rv40_chroma_mc4_mmx;
60 c->rv40_weight_pixels_tab[0] = ff_rv40_weight_func_16_mmx;
61 c->rv40_weight_pixels_tab[1] = ff_rv40_weight_func_8_mmx;
63 if (mm_flags & AV_CPU_FLAG_MMX2) {
64 c->avg_chroma_pixels_tab[0] = ff_avg_rv40_chroma_mc8_mmx2;
65 c->avg_chroma_pixels_tab[1] = ff_avg_rv40_chroma_mc4_mmx2;
66 } else if (mm_flags & AV_CPU_FLAG_3DNOW) {
67 c->avg_chroma_pixels_tab[0] = ff_avg_rv40_chroma_mc8_3dnow;
68 c->avg_chroma_pixels_tab[1] = ff_avg_rv40_chroma_mc4_3dnow;
70 if (mm_flags & AV_CPU_FLAG_SSE2) {
71 c->rv40_weight_pixels_tab[0] = ff_rv40_weight_func_16_sse2;
72 c->rv40_weight_pixels_tab[1] = ff_rv40_weight_func_8_sse2;
74 if (mm_flags & AV_CPU_FLAG_SSSE3) {
75 c->rv40_weight_pixels_tab[0] = ff_rv40_weight_func_16_ssse3;
76 c->rv40_weight_pixels_tab[1] = ff_rv40_weight_func_8_ssse3;