2 * ARM NEON optimised DSP functions
3 * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
5 * This file is part of FFmpeg.
7 * FFmpeg 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 * FFmpeg 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 FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
25 #include "libavutil/attributes.h"
26 #include "hpeldsp_arm.h"
28 void ff_put_pixels16_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
29 void ff_put_pixels16_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
30 void ff_put_pixels16_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
31 void ff_put_pixels16_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
32 void ff_put_pixels8_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
33 void ff_put_pixels8_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
34 void ff_put_pixels8_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
35 void ff_put_pixels8_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
36 void ff_put_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
37 void ff_put_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
38 void ff_put_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
39 void ff_put_pixels8_x2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
40 void ff_put_pixels8_y2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
41 void ff_put_pixels8_xy2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
43 void ff_avg_pixels16_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
44 void ff_avg_pixels16_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
45 void ff_avg_pixels16_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
46 void ff_avg_pixels16_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
47 void ff_avg_pixels8_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
48 void ff_avg_pixels8_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
49 void ff_avg_pixels8_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
50 void ff_avg_pixels8_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
51 void ff_avg_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
52 void ff_avg_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
53 void ff_avg_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
55 av_cold
void ff_hpeldsp_init_neon(HpelDSPContext
*c
, int flags
)
57 c
->put_pixels_tab
[0][0] = ff_put_pixels16_neon
;
58 c
->put_pixels_tab
[0][1] = ff_put_pixels16_x2_neon
;
59 c
->put_pixels_tab
[0][2] = ff_put_pixels16_y2_neon
;
60 c
->put_pixels_tab
[0][3] = ff_put_pixels16_xy2_neon
;
61 c
->put_pixels_tab
[1][0] = ff_put_pixels8_neon
;
62 c
->put_pixels_tab
[1][1] = ff_put_pixels8_x2_neon
;
63 c
->put_pixels_tab
[1][2] = ff_put_pixels8_y2_neon
;
64 c
->put_pixels_tab
[1][3] = ff_put_pixels8_xy2_neon
;
66 c
->put_no_rnd_pixels_tab
[0][0] = ff_put_pixels16_neon
;
67 c
->put_no_rnd_pixels_tab
[0][1] = ff_put_pixels16_x2_no_rnd_neon
;
68 c
->put_no_rnd_pixels_tab
[0][2] = ff_put_pixels16_y2_no_rnd_neon
;
69 c
->put_no_rnd_pixels_tab
[0][3] = ff_put_pixels16_xy2_no_rnd_neon
;
70 c
->put_no_rnd_pixels_tab
[1][0] = ff_put_pixels8_neon
;
71 c
->put_no_rnd_pixels_tab
[1][1] = ff_put_pixels8_x2_no_rnd_neon
;
72 c
->put_no_rnd_pixels_tab
[1][2] = ff_put_pixels8_y2_no_rnd_neon
;
73 c
->put_no_rnd_pixels_tab
[1][3] = ff_put_pixels8_xy2_no_rnd_neon
;
75 c
->avg_pixels_tab
[0][0] = ff_avg_pixels16_neon
;
76 c
->avg_pixels_tab
[0][1] = ff_avg_pixels16_x2_neon
;
77 c
->avg_pixels_tab
[0][2] = ff_avg_pixels16_y2_neon
;
78 c
->avg_pixels_tab
[0][3] = ff_avg_pixels16_xy2_neon
;
79 c
->avg_pixels_tab
[1][0] = ff_avg_pixels8_neon
;
80 c
->avg_pixels_tab
[1][1] = ff_avg_pixels8_x2_neon
;
81 c
->avg_pixels_tab
[1][2] = ff_avg_pixels8_y2_neon
;
82 c
->avg_pixels_tab
[1][3] = ff_avg_pixels8_xy2_neon
;
84 c
->avg_no_rnd_pixels_tab
[0] = ff_avg_pixels16_neon
;
85 c
->avg_no_rnd_pixels_tab
[1] = ff_avg_pixels16_x2_no_rnd_neon
;
86 c
->avg_no_rnd_pixels_tab
[2] = ff_avg_pixels16_y2_no_rnd_neon
;
87 c
->avg_no_rnd_pixels_tab
[3] = ff_avg_pixels16_xy2_no_rnd_neon
;