FFmpeg
h264dsp_init_aarch64.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2010 Mans Rullgard <mans@mansr.com>
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg 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  * FFmpeg 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 FFmpeg; 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/cpu.h"
25 #include "libavutil/aarch64/cpu.h"
26 #include "libavcodec/h264dsp.h"
27 
28 void ff_h264_v_loop_filter_luma_neon(uint8_t *pix, ptrdiff_t stride, int alpha,
29  int beta, int8_t *tc0);
30 void ff_h264_h_loop_filter_luma_neon(uint8_t *pix, ptrdiff_t stride, int alpha,
31  int beta, int8_t *tc0);
32 void ff_h264_v_loop_filter_luma_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha,
33  int beta);
34 void ff_h264_h_loop_filter_luma_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha,
35  int beta);
36 void ff_h264_v_loop_filter_chroma_neon(uint8_t *pix, ptrdiff_t stride, int alpha,
37  int beta, int8_t *tc0);
38 void ff_h264_h_loop_filter_chroma_neon(uint8_t *pix, ptrdiff_t stride, int alpha,
39  int beta, int8_t *tc0);
40 void ff_h264_h_loop_filter_chroma422_neon(uint8_t *pix, ptrdiff_t stride, int alpha,
41  int beta, int8_t *tc0);
42 void ff_h264_v_loop_filter_chroma_intra_neon(uint8_t *pix, ptrdiff_t stride,
43  int alpha, int beta);
44 void ff_h264_h_loop_filter_chroma_intra_neon(uint8_t *pix, ptrdiff_t stride,
45  int alpha, int beta);
46 void ff_h264_h_loop_filter_chroma422_intra_neon(uint8_t *pix, ptrdiff_t stride,
47  int alpha, int beta);
48 void ff_h264_h_loop_filter_chroma_mbaff_intra_neon(uint8_t *pix, ptrdiff_t stride,
49  int alpha, int beta);
50 
51 void ff_weight_h264_pixels_16_neon(uint8_t *dst, ptrdiff_t stride, int height,
52  int log2_den, int weight, int offset);
53 void ff_weight_h264_pixels_8_neon(uint8_t *dst, ptrdiff_t stride, int height,
54  int log2_den, int weight, int offset);
55 void ff_weight_h264_pixels_4_neon(uint8_t *dst, ptrdiff_t stride, int height,
56  int log2_den, int weight, int offset);
57 
58 void ff_biweight_h264_pixels_16_neon(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
59  int height, int log2_den, int weightd,
60  int weights, int offset);
61 void ff_biweight_h264_pixels_8_neon(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
62  int height, int log2_den, int weightd,
63  int weights, int offset);
64 void ff_biweight_h264_pixels_4_neon(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
65  int height, int log2_den, int weightd,
66  int weights, int offset);
67 
68 void ff_h264_idct_add_neon(uint8_t *dst, int16_t *block, int stride);
69 void ff_h264_idct_dc_add_neon(uint8_t *dst, int16_t *block, int stride);
70 void ff_h264_idct_add16_neon(uint8_t *dst, const int *block_offset,
71  int16_t *block, int stride,
72  const uint8_t nnzc[5 * 8]);
73 void ff_h264_idct_add16intra_neon(uint8_t *dst, const int *block_offset,
74  int16_t *block, int stride,
75  const uint8_t nnzc[5 * 8]);
76 void ff_h264_idct_add8_neon(uint8_t **dest, const int *block_offset,
77  int16_t *block, int stride,
78  const uint8_t nnzc[15 * 8]);
79 
80 void ff_h264_idct8_add_neon(uint8_t *dst, int16_t *block, int stride);
81 void ff_h264_idct8_dc_add_neon(uint8_t *dst, int16_t *block, int stride);
82 void ff_h264_idct8_add4_neon(uint8_t *dst, const int *block_offset,
83  int16_t *block, int stride,
84  const uint8_t nnzc[5 * 8]);
85 
86 void ff_h264_v_loop_filter_luma_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha,
87  int beta, int8_t *tc0);
88 void ff_h264_h_loop_filter_luma_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha,
89  int beta, int8_t *tc0);
90 void ff_h264_v_loop_filter_luma_intra_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha,
91  int beta);
92 void ff_h264_h_loop_filter_luma_intra_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha,
93  int beta);
94 void ff_h264_v_loop_filter_chroma_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha,
95  int beta, int8_t *tc0);
96 void ff_h264_h_loop_filter_chroma_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha,
97  int beta, int8_t *tc0);
98 void ff_h264_h_loop_filter_chroma422_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha,
99  int beta, int8_t *tc0);
100 void ff_h264_v_loop_filter_chroma_intra_neon_10(uint8_t *pix, ptrdiff_t stride,
101  int alpha, int beta);
102 void ff_h264_h_loop_filter_chroma_intra_neon_10(uint8_t *pix, ptrdiff_t stride,
103  int alpha, int beta);
104 void ff_h264_h_loop_filter_chroma422_intra_neon_10(uint8_t *pix, ptrdiff_t stride,
105  int alpha, int beta);
106 void ff_h264_h_loop_filter_chroma_mbaff_intra_neon_10(uint8_t *pix, ptrdiff_t stride,
107  int alpha, int beta);
108 
110  const int chroma_format_idc)
111 {
112  int cpu_flags = av_get_cpu_flags();
113 
114  if (have_neon(cpu_flags) && bit_depth == 8) {
115  c->h264_v_loop_filter_luma = ff_h264_v_loop_filter_luma_neon;
116  c->h264_h_loop_filter_luma = ff_h264_h_loop_filter_luma_neon;
117  c->h264_v_loop_filter_luma_intra= ff_h264_v_loop_filter_luma_intra_neon;
118  c->h264_h_loop_filter_luma_intra= ff_h264_h_loop_filter_luma_intra_neon;
119 
120  c->h264_v_loop_filter_chroma = ff_h264_v_loop_filter_chroma_neon;
121  c->h264_v_loop_filter_chroma_intra = ff_h264_v_loop_filter_chroma_intra_neon;
122 
123  if (chroma_format_idc <= 1) {
124  c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma_neon;
125  c->h264_h_loop_filter_chroma_intra = ff_h264_h_loop_filter_chroma_intra_neon;
126  c->h264_h_loop_filter_chroma_mbaff_intra = ff_h264_h_loop_filter_chroma_mbaff_intra_neon;
127  } else {
128  c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma422_neon;
129  c->h264_h_loop_filter_chroma_mbaff = ff_h264_h_loop_filter_chroma_neon;
130  c->h264_h_loop_filter_chroma_intra = ff_h264_h_loop_filter_chroma422_intra_neon;
131  c->h264_h_loop_filter_chroma_mbaff_intra = ff_h264_h_loop_filter_chroma_intra_neon;
132  }
133 
134  c->weight_h264_pixels_tab[0] = ff_weight_h264_pixels_16_neon;
135  c->weight_h264_pixels_tab[1] = ff_weight_h264_pixels_8_neon;
136  c->weight_h264_pixels_tab[2] = ff_weight_h264_pixels_4_neon;
137 
138  c->biweight_h264_pixels_tab[0] = ff_biweight_h264_pixels_16_neon;
139  c->biweight_h264_pixels_tab[1] = ff_biweight_h264_pixels_8_neon;
140  c->biweight_h264_pixels_tab[2] = ff_biweight_h264_pixels_4_neon;
141 
142  c->h264_idct_add = ff_h264_idct_add_neon;
143  c->h264_idct_dc_add = ff_h264_idct_dc_add_neon;
144  c->h264_idct_add16 = ff_h264_idct_add16_neon;
145  c->h264_idct_add16intra = ff_h264_idct_add16intra_neon;
146  if (chroma_format_idc <= 1)
147  c->h264_idct_add8 = ff_h264_idct_add8_neon;
148  c->h264_idct8_add = ff_h264_idct8_add_neon;
149  c->h264_idct8_dc_add = ff_h264_idct8_dc_add_neon;
150  c->h264_idct8_add4 = ff_h264_idct8_add4_neon;
151  } else if (have_neon(cpu_flags) && bit_depth == 10) {
152  c->h264_v_loop_filter_chroma = ff_h264_v_loop_filter_chroma_neon_10;
153  c->h264_v_loop_filter_chroma_intra = ff_h264_v_loop_filter_chroma_intra_neon_10;
154 
155  if (chroma_format_idc <= 1) {
156  c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma_neon_10;
157  c->h264_h_loop_filter_chroma_intra = ff_h264_h_loop_filter_chroma_intra_neon_10;
158  c->h264_h_loop_filter_chroma_mbaff_intra = ff_h264_h_loop_filter_chroma_mbaff_intra_neon_10;
159  } else {
160  c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma422_neon_10;
161  c->h264_h_loop_filter_chroma_mbaff = ff_h264_h_loop_filter_chroma_neon_10;
162  c->h264_h_loop_filter_chroma_intra = ff_h264_h_loop_filter_chroma422_intra_neon_10;
163  c->h264_h_loop_filter_chroma_mbaff_intra = ff_h264_h_loop_filter_chroma_intra_neon_10;
164  }
165  }
166 }
ff_h264_idct8_add4_neon
void ff_h264_idct8_add4_neon(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[5 *8])
ff_biweight_h264_pixels_4_neon
void ff_biweight_h264_pixels_4_neon(uint8_t *dst, uint8_t *src, ptrdiff_t stride, int height, int log2_den, int weightd, int weights, int offset)
ff_h264_v_loop_filter_luma_neon
void ff_h264_v_loop_filter_luma_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
ff_h264_idct_add8_neon
void ff_h264_idct_add8_neon(uint8_t **dest, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15 *8])
ff_weight_h264_pixels_8_neon
void ff_weight_h264_pixels_8_neon(uint8_t *dst, ptrdiff_t stride, int height, int log2_den, int weight, int offset)
ff_h264_h_loop_filter_luma_intra_neon
void ff_h264_h_loop_filter_luma_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
ff_h264_idct8_dc_add_neon
void ff_h264_idct8_dc_add_neon(uint8_t *dst, int16_t *block, int stride)
av_get_cpu_flags
int av_get_cpu_flags(void)
Return the flags which specify extensions supported by the CPU.
Definition: cpu.c:103
ff_h264_idct_dc_add_neon
void ff_h264_idct_dc_add_neon(uint8_t *dst, int16_t *block, int stride)
ff_h264_idct_add16intra_neon
void ff_h264_idct_add16intra_neon(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[5 *8])
bit_depth
static void bit_depth(AudioStatsContext *s, const uint64_t *const mask, uint8_t *depth)
Definition: af_astats.c:245
cpu_flags
static atomic_int cpu_flags
Definition: cpu.c:52
ff_h264_v_loop_filter_chroma_neon
void ff_h264_v_loop_filter_chroma_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
ff_h264_h_loop_filter_chroma422_neon
void ff_h264_h_loop_filter_chroma422_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
ff_biweight_h264_pixels_16_neon
void ff_biweight_h264_pixels_16_neon(uint8_t *dst, uint8_t *src, ptrdiff_t stride, int height, int log2_den, int weightd, int weights, int offset)
ff_h264_h_loop_filter_chroma_mbaff_intra_neon
void ff_h264_h_loop_filter_chroma_mbaff_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
av_cold
#define av_cold
Definition: attributes.h:90
ff_weight_h264_pixels_16_neon
void ff_weight_h264_pixels_16_neon(uint8_t *dst, ptrdiff_t stride, int height, int log2_den, int weight, int offset)
ff_h264_idct_add16_neon
void ff_h264_idct_add16_neon(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[5 *8])
ff_h264_idct_add_neon
void ff_h264_idct_add_neon(uint8_t *dst, int16_t *block, int stride)
ff_h264_v_loop_filter_chroma_intra_neon
void ff_h264_v_loop_filter_chroma_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
ff_h264_h_loop_filter_chroma_neon
void ff_h264_h_loop_filter_chroma_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
h264dsp.h
c
Undefined Behavior In the C some operations are like signed integer dereferencing freed accessing outside allocated Undefined Behavior must not occur in a C it is not safe even if the output of undefined operations is unused The unsafety may seem nit picking but Optimizing compilers have in fact optimized code on the assumption that no undefined Behavior occurs Optimizing code based on wrong assumptions can and has in some cases lead to effects beyond the output of computations The signed integer overflow problem in speed critical code Code which is highly optimized and works with signed integers sometimes has the problem that often the output of the computation does not c
Definition: undefined.txt:32
weight
static int weight(int i, int blen, int offset)
Definition: diracdec.c:1562
H264DSPContext
Context for storing H.264 DSP functions.
Definition: h264dsp.h:42
cpu.h
ff_h264_v_loop_filter_luma_intra_neon_10
void ff_h264_v_loop_filter_luma_intra_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
have_neon
#define have_neon(flags)
Definition: cpu.h:26
ff_h264_h_loop_filter_luma_neon_10
void ff_h264_h_loop_filter_luma_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
height
#define height
offset
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf offset
Definition: writing_filters.txt:86
attributes.h
ff_h264_h_loop_filter_chroma_mbaff_intra_neon_10
void ff_h264_h_loop_filter_chroma_mbaff_intra_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
ff_h264_idct8_add_neon
void ff_h264_idct8_add_neon(uint8_t *dst, int16_t *block, int stride)
weights
static const int weights[]
Definition: hevc_pel.c:32
ff_h264_h_loop_filter_chroma422_intra_neon_10
void ff_h264_h_loop_filter_chroma422_intra_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
ff_h264_h_loop_filter_luma_neon
void ff_h264_h_loop_filter_luma_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
ff_h264_h_loop_filter_chroma422_neon_10
void ff_h264_h_loop_filter_chroma422_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
ff_h264_h_loop_filter_luma_intra_neon_10
void ff_h264_h_loop_filter_luma_intra_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
ff_h264_v_loop_filter_luma_neon_10
void ff_h264_v_loop_filter_luma_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
stride
#define stride
Definition: h264pred_template.c:537
ff_h264_v_loop_filter_chroma_intra_neon_10
void ff_h264_v_loop_filter_chroma_intra_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
ff_h264_h_loop_filter_chroma_intra_neon
void ff_h264_h_loop_filter_chroma_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
ff_h264_h_loop_filter_chroma_intra_neon_10
void ff_h264_h_loop_filter_chroma_intra_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
ff_h264_h_loop_filter_chroma_neon_10
void ff_h264_h_loop_filter_chroma_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
ff_h264_v_loop_filter_chroma_neon_10
void ff_h264_v_loop_filter_chroma_neon_10(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
ff_h264_h_loop_filter_chroma422_intra_neon
void ff_h264_h_loop_filter_chroma422_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
ff_biweight_h264_pixels_8_neon
void ff_biweight_h264_pixels_8_neon(uint8_t *dst, uint8_t *src, ptrdiff_t stride, int height, int log2_den, int weightd, int weights, int offset)
ff_h264_v_loop_filter_luma_intra_neon
void ff_h264_v_loop_filter_luma_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
alpha
static const int16_t alpha[]
Definition: ilbcdata.h:55
ff_h264dsp_init_aarch64
av_cold void ff_h264dsp_init_aarch64(H264DSPContext *c, const int bit_depth, const int chroma_format_idc)
Definition: h264dsp_init_aarch64.c:109
src
INIT_CLIP pixel * src
Definition: h264pred_template.c:418
block
The exact code depends on how similar the blocks are and how related they are to the block
Definition: filter_design.txt:207
cpu.h
ff_weight_h264_pixels_4_neon
void ff_weight_h264_pixels_4_neon(uint8_t *dst, ptrdiff_t stride, int height, int log2_den, int weight, int offset)