arm: Implement a NEON version of 422 h264_h_loop_filter_chroma
[libav.git] / libavcodec / arm / h264dsp_init_arm.c
1 /*
2 * Copyright (c) 2010 Mans Rullgard <mans@mansr.com>
3 *
4 * This file is part of Libav.
5 *
6 * Libav is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
10 *
11 * Libav is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with Libav; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 */
20
21 #include <stdint.h>
22
23 #include "libavutil/attributes.h"
24 #include "libavutil/arm/cpu.h"
25 #include "libavcodec/h264dsp.h"
26 #include "libavcodec/arm/startcode.h"
27
28 void ff_h264_v_loop_filter_luma_neon(uint8_t *pix, int stride, int alpha,
29 int beta, int8_t *tc0);
30 void ff_h264_h_loop_filter_luma_neon(uint8_t *pix, int stride, int alpha,
31 int beta, int8_t *tc0);
32 void ff_h264_v_loop_filter_chroma_neon(uint8_t *pix, int stride, int alpha,
33 int beta, int8_t *tc0);
34 void ff_h264_h_loop_filter_chroma_neon(uint8_t *pix, int stride, int alpha,
35 int beta, int8_t *tc0);
36 void ff_h264_h_loop_filter_chroma422_neon(uint8_t *pix, int stride, int alpha,
37 int beta, int8_t *tc0);
38
39 void ff_weight_h264_pixels_16_neon(uint8_t *dst, int stride, int height,
40 int log2_den, int weight, int offset);
41 void ff_weight_h264_pixels_8_neon(uint8_t *dst, int stride, int height,
42 int log2_den, int weight, int offset);
43 void ff_weight_h264_pixels_4_neon(uint8_t *dst, int stride, int height,
44 int log2_den, int weight, int offset);
45
46 void ff_biweight_h264_pixels_16_neon(uint8_t *dst, uint8_t *src, int stride,
47 int height, int log2_den, int weightd,
48 int weights, int offset);
49 void ff_biweight_h264_pixels_8_neon(uint8_t *dst, uint8_t *src, int stride,
50 int height, int log2_den, int weightd,
51 int weights, int offset);
52 void ff_biweight_h264_pixels_4_neon(uint8_t *dst, uint8_t *src, int stride,
53 int height, int log2_den, int weightd,
54 int weights, int offset);
55
56 void ff_h264_idct_add_neon(uint8_t *dst, int16_t *block, int stride);
57 void ff_h264_idct_dc_add_neon(uint8_t *dst, int16_t *block, int stride);
58 void ff_h264_idct_add16_neon(uint8_t *dst, const int *block_offset,
59 int16_t *block, int stride,
60 const uint8_t nnzc[6*8]);
61 void ff_h264_idct_add16intra_neon(uint8_t *dst, const int *block_offset,
62 int16_t *block, int stride,
63 const uint8_t nnzc[6*8]);
64 void ff_h264_idct_add8_neon(uint8_t **dest, const int *block_offset,
65 int16_t *block, int stride,
66 const uint8_t nnzc[6*8]);
67
68 void ff_h264_idct8_add_neon(uint8_t *dst, int16_t *block, int stride);
69 void ff_h264_idct8_dc_add_neon(uint8_t *dst, int16_t *block, int stride);
70 void ff_h264_idct8_add4_neon(uint8_t *dst, const int *block_offset,
71 int16_t *block, int stride,
72 const uint8_t nnzc[6*8]);
73
74 static av_cold void h264dsp_init_neon(H264DSPContext *c, const int bit_depth,
75 const int chroma_format_idc)
76 {
77 if (bit_depth == 8) {
78 c->h264_v_loop_filter_luma = ff_h264_v_loop_filter_luma_neon;
79 c->h264_h_loop_filter_luma = ff_h264_h_loop_filter_luma_neon;
80 c->h264_v_loop_filter_chroma = ff_h264_v_loop_filter_chroma_neon;
81
82 if (chroma_format_idc <= 1)
83 c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma_neon;
84 else
85 c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma422_neon;
86
87 c->weight_h264_pixels_tab[0] = ff_weight_h264_pixels_16_neon;
88 c->weight_h264_pixels_tab[1] = ff_weight_h264_pixels_8_neon;
89 c->weight_h264_pixels_tab[2] = ff_weight_h264_pixels_4_neon;
90
91 c->biweight_h264_pixels_tab[0] = ff_biweight_h264_pixels_16_neon;
92 c->biweight_h264_pixels_tab[1] = ff_biweight_h264_pixels_8_neon;
93 c->biweight_h264_pixels_tab[2] = ff_biweight_h264_pixels_4_neon;
94
95 c->h264_idct_add = ff_h264_idct_add_neon;
96 c->h264_idct_dc_add = ff_h264_idct_dc_add_neon;
97 c->h264_idct_add16 = ff_h264_idct_add16_neon;
98 c->h264_idct_add16intra = ff_h264_idct_add16intra_neon;
99 if (chroma_format_idc <= 1)
100 c->h264_idct_add8 = ff_h264_idct_add8_neon;
101 c->h264_idct8_add = ff_h264_idct8_add_neon;
102 c->h264_idct8_dc_add = ff_h264_idct8_dc_add_neon;
103 c->h264_idct8_add4 = ff_h264_idct8_add4_neon;
104 }
105 }
106
107 av_cold void ff_h264dsp_init_arm(H264DSPContext *c, const int bit_depth,
108 const int chroma_format_idc)
109 {
110 int cpu_flags = av_get_cpu_flags();
111
112 if (have_setend(cpu_flags))
113 c->startcode_find_candidate = ff_startcode_find_candidate_armv6;
114 if (have_neon(cpu_flags))
115 h264dsp_init_neon(c, bit_depth, chroma_format_idc);
116 }