FFmpeg  4.4.4
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 
29  int beta, int8_t *tc0);
31  int beta, int8_t *tc0);
33  int beta);
35  int beta);
37  int beta, int8_t *tc0);
39  int beta, int8_t *tc0);
41  int beta, int8_t *tc0);
43  int alpha, int beta);
45  int alpha, int beta);
47  int alpha, int beta);
49  int alpha, int beta);
50 
52  int log2_den, int weight, int offset);
54  int log2_den, int weight, int offset);
56  int log2_den, int weight, int offset);
57 
59  int height, int log2_den, int weightd,
60  int weights, int offset);
62  int height, int log2_den, int weightd,
63  int weights, int offset);
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[6*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[6*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[6*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[6*8]);
85 
87  const int chroma_format_idc)
88 {
90 
91  if (have_neon(cpu_flags) && bit_depth == 8) {
92  c->h264_v_loop_filter_luma = ff_h264_v_loop_filter_luma_neon;
93  c->h264_h_loop_filter_luma = ff_h264_h_loop_filter_luma_neon;
94  c->h264_v_loop_filter_luma_intra= ff_h264_v_loop_filter_luma_intra_neon;
95  c->h264_h_loop_filter_luma_intra= ff_h264_h_loop_filter_luma_intra_neon;
96 
97  c->h264_v_loop_filter_chroma = ff_h264_v_loop_filter_chroma_neon;
98  c->h264_v_loop_filter_chroma_intra = ff_h264_v_loop_filter_chroma_intra_neon;
99 
100  if (chroma_format_idc <= 1) {
101  c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma_neon;
102  c->h264_h_loop_filter_chroma_intra = ff_h264_h_loop_filter_chroma_intra_neon;
103  c->h264_h_loop_filter_chroma_mbaff_intra = ff_h264_h_loop_filter_chroma_mbaff_intra_neon;
104  } else {
105  c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma422_neon;
106  c->h264_h_loop_filter_chroma_mbaff = ff_h264_h_loop_filter_chroma_neon;
107  c->h264_h_loop_filter_chroma_intra = ff_h264_h_loop_filter_chroma422_intra_neon;
108  c->h264_h_loop_filter_chroma_mbaff_intra = ff_h264_h_loop_filter_chroma_intra_neon;
109  }
110 
111  c->weight_h264_pixels_tab[0] = ff_weight_h264_pixels_16_neon;
112  c->weight_h264_pixels_tab[1] = ff_weight_h264_pixels_8_neon;
113  c->weight_h264_pixels_tab[2] = ff_weight_h264_pixels_4_neon;
114 
115  c->biweight_h264_pixels_tab[0] = ff_biweight_h264_pixels_16_neon;
116  c->biweight_h264_pixels_tab[1] = ff_biweight_h264_pixels_8_neon;
117  c->biweight_h264_pixels_tab[2] = ff_biweight_h264_pixels_4_neon;
118 
119  c->h264_idct_add = ff_h264_idct_add_neon;
120  c->h264_idct_dc_add = ff_h264_idct_dc_add_neon;
121  c->h264_idct_add16 = ff_h264_idct_add16_neon;
122  c->h264_idct_add16intra = ff_h264_idct_add16intra_neon;
123  if (chroma_format_idc <= 1)
124  c->h264_idct_add8 = ff_h264_idct_add8_neon;
125  c->h264_idct8_add = ff_h264_idct8_add_neon;
126  c->h264_idct8_dc_add = ff_h264_idct8_dc_add_neon;
127  c->h264_idct8_add4 = ff_h264_idct8_add4_neon;
128  }
129 }
#define have_neon(flags)
Definition: cpu.h:26
static void bit_depth(AudioStatsContext *s, uint64_t mask, uint64_t imask, AVRational *depth)
Definition: af_astats.c:254
Macro definitions for various function/variable attributes.
#define av_cold
Definition: attributes.h:88
uint8_t
static atomic_int cpu_flags
Definition: cpu.c:50
int av_get_cpu_flags(void)
Return the flags which specify extensions supported by the CPU.
Definition: cpu.c:95
H.264 DSP functions.
void ff_weight_h264_pixels_4_neon(uint8_t *dst, ptrdiff_t stride, int height, int log2_den, int weight, int offset)
void ff_h264_h_loop_filter_chroma_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
void ff_h264_idct_add16_neon(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[6 *8])
void ff_h264_idct_add16intra_neon(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[6 *8])
void ff_h264_idct8_dc_add_neon(uint8_t *dst, int16_t *block, int stride)
void ff_h264_v_loop_filter_luma_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
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)
av_cold void ff_h264dsp_init_aarch64(H264DSPContext *c, const int bit_depth, const int chroma_format_idc)
void ff_h264_idct_add8_neon(uint8_t **dest, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[6 *8])
void ff_h264_idct8_add4_neon(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[6 *8])
void ff_h264_h_loop_filter_chroma_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
void ff_h264_h_loop_filter_chroma422_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
void ff_h264_h_loop_filter_luma_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
void ff_h264_idct_dc_add_neon(uint8_t *dst, int16_t *block, int stride)
void ff_h264_idct_add_neon(uint8_t *dst, int16_t *block, int stride)
void ff_h264_v_loop_filter_chroma_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
void ff_weight_h264_pixels_16_neon(uint8_t *dst, ptrdiff_t stride, int height, int log2_den, int weight, int offset)
void ff_h264_v_loop_filter_chroma_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
void ff_h264_h_loop_filter_luma_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
void ff_weight_h264_pixels_8_neon(uint8_t *dst, ptrdiff_t stride, int height, int log2_den, int weight, int offset)
void ff_h264_h_loop_filter_chroma_mbaff_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
void ff_h264_idct8_add_neon(uint8_t *dst, int16_t *block, int stride)
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)
void ff_h264_h_loop_filter_chroma422_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0)
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)
void ff_h264_v_loop_filter_luma_intra_neon(uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
static const int weights[]
Definition: hevc_pel.c:32
static const int16_t alpha[]
Definition: ilbcdata.h:55
static int weight(int i, int blen, int offset)
Definition: diracdec.c:1561
int stride
Definition: mace.c:144
Context for storing H.264 DSP functions.
Definition: h264dsp.h:42
#define src
Definition: vp8dsp.c:255
static int16_t block[64]
Definition: dct.c:116
#define height
static const uint8_t offset[127][2]
Definition: vf_spp.c:107
static double c[64]