FFmpeg
hevcdsp_init_aarch64.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2020 Reimar Döffinger
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/hevcdsp.h"
27 
28 void ff_hevc_add_residual_4x4_8_neon(uint8_t *_dst, int16_t *coeffs,
29  ptrdiff_t stride);
30 void ff_hevc_add_residual_4x4_10_neon(uint8_t *_dst, int16_t *coeffs,
31  ptrdiff_t stride);
32 void ff_hevc_add_residual_8x8_8_neon(uint8_t *_dst, int16_t *coeffs,
33  ptrdiff_t stride);
34 void ff_hevc_add_residual_8x8_10_neon(uint8_t *_dst, int16_t *coeffs,
35  ptrdiff_t stride);
36 void ff_hevc_add_residual_16x16_8_neon(uint8_t *_dst, int16_t *coeffs,
37  ptrdiff_t stride);
38 void ff_hevc_add_residual_16x16_10_neon(uint8_t *_dst, int16_t *coeffs,
39  ptrdiff_t stride);
40 void ff_hevc_add_residual_32x32_8_neon(uint8_t *_dst, int16_t *coeffs,
41  ptrdiff_t stride);
42 void ff_hevc_add_residual_32x32_10_neon(uint8_t *_dst, int16_t *coeffs,
43  ptrdiff_t stride);
44 void ff_hevc_idct_8x8_8_neon(int16_t *coeffs, int col_limit);
45 void ff_hevc_idct_8x8_10_neon(int16_t *coeffs, int col_limit);
46 void ff_hevc_idct_16x16_8_neon(int16_t *coeffs, int col_limit);
47 void ff_hevc_idct_16x16_10_neon(int16_t *coeffs, int col_limit);
48 void ff_hevc_idct_4x4_dc_8_neon(int16_t *coeffs);
49 void ff_hevc_idct_8x8_dc_8_neon(int16_t *coeffs);
50 void ff_hevc_idct_16x16_dc_8_neon(int16_t *coeffs);
51 void ff_hevc_idct_32x32_dc_8_neon(int16_t *coeffs);
52 void ff_hevc_idct_4x4_dc_10_neon(int16_t *coeffs);
53 void ff_hevc_idct_8x8_dc_10_neon(int16_t *coeffs);
54 void ff_hevc_idct_16x16_dc_10_neon(int16_t *coeffs);
55 void ff_hevc_idct_32x32_dc_10_neon(int16_t *coeffs);
56 void ff_hevc_sao_band_filter_8x8_8_neon(uint8_t *_dst, uint8_t *_src,
57  ptrdiff_t stride_dst, ptrdiff_t stride_src,
58  int16_t *sao_offset_val, int sao_left_class,
59  int width, int height);
60 
61 
62 
64 {
65  if (!have_neon(av_get_cpu_flags())) return;
66 
67  if (bit_depth == 8) {
68  c->add_residual[0] = ff_hevc_add_residual_4x4_8_neon;
69  c->add_residual[1] = ff_hevc_add_residual_8x8_8_neon;
70  c->add_residual[2] = ff_hevc_add_residual_16x16_8_neon;
71  c->add_residual[3] = ff_hevc_add_residual_32x32_8_neon;
72  c->idct[1] = ff_hevc_idct_8x8_8_neon;
73  c->idct[2] = ff_hevc_idct_16x16_8_neon;
74  c->idct_dc[0] = ff_hevc_idct_4x4_dc_8_neon;
75  c->idct_dc[1] = ff_hevc_idct_8x8_dc_8_neon;
76  c->idct_dc[2] = ff_hevc_idct_16x16_dc_8_neon;
77  c->idct_dc[3] = ff_hevc_idct_32x32_dc_8_neon;
78  // This function is disabled, as it doesn't handle widths that aren't
79  // an even multiple of 8 correctly. fate-hevc doesn't exercise that
80  // for the current size, but if enabled for bigger sizes, the cases
81  // of non-multiple of 8 seem to arise.
82 // c->sao_band_filter[0] = ff_hevc_sao_band_filter_8x8_8_neon;
83  }
84  if (bit_depth == 10) {
85  c->add_residual[0] = ff_hevc_add_residual_4x4_10_neon;
86  c->add_residual[1] = ff_hevc_add_residual_8x8_10_neon;
87  c->add_residual[2] = ff_hevc_add_residual_16x16_10_neon;
88  c->add_residual[3] = ff_hevc_add_residual_32x32_10_neon;
89  c->idct[1] = ff_hevc_idct_8x8_10_neon;
90  c->idct[2] = ff_hevc_idct_16x16_10_neon;
91  c->idct_dc[0] = ff_hevc_idct_4x4_dc_10_neon;
92  c->idct_dc[1] = ff_hevc_idct_8x8_dc_10_neon;
93  c->idct_dc[2] = ff_hevc_idct_16x16_dc_10_neon;
94  c->idct_dc[3] = ff_hevc_idct_32x32_dc_10_neon;
95  }
96 }
stride
int stride
Definition: mace.c:144
bit_depth
static void bit_depth(AudioStatsContext *s, uint64_t mask, uint64_t imask, AVRational *depth)
Definition: af_astats.c:226
ff_hevc_idct_4x4_dc_8_neon
void ff_hevc_idct_4x4_dc_8_neon(int16_t *coeffs)
ff_hevc_idct_16x16_8_neon
void ff_hevc_idct_16x16_8_neon(int16_t *coeffs, int col_limit)
av_get_cpu_flags
int av_get_cpu_flags(void)
Return the flags which specify extensions supported by the CPU.
Definition: cpu.c:98
ff_hevc_idct_8x8_8_neon
void ff_hevc_idct_8x8_8_neon(int16_t *coeffs, int col_limit)
ff_hevc_idct_16x16_dc_10_neon
void ff_hevc_idct_16x16_dc_10_neon(int16_t *coeffs)
av_cold
#define av_cold
Definition: attributes.h:90
width
#define width
hevcdsp.h
ff_hevc_add_residual_32x32_8_neon
void ff_hevc_add_residual_32x32_8_neon(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride)
ff_hevc_idct_16x16_dc_8_neon
void ff_hevc_idct_16x16_dc_8_neon(int16_t *coeffs)
ff_hevc_idct_32x32_dc_10_neon
void ff_hevc_idct_32x32_dc_10_neon(int16_t *coeffs)
ff_hevc_add_residual_8x8_8_neon
void ff_hevc_add_residual_8x8_8_neon(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride)
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
ff_hevc_add_residual_4x4_10_neon
void ff_hevc_add_residual_4x4_10_neon(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride)
cpu.h
ff_hevc_idct_4x4_dc_10_neon
void ff_hevc_idct_4x4_dc_10_neon(int16_t *coeffs)
have_neon
#define have_neon(flags)
Definition: cpu.h:26
height
#define height
ff_hevc_idct_8x8_10_neon
void ff_hevc_idct_8x8_10_neon(int16_t *coeffs, int col_limit)
HEVCDSPContext
Definition: hevcdsp.h:47
attributes.h
ff_hevc_idct_32x32_dc_8_neon
void ff_hevc_idct_32x32_dc_8_neon(int16_t *coeffs)
ff_hevc_idct_8x8_dc_8_neon
void ff_hevc_idct_8x8_dc_8_neon(int16_t *coeffs)
ff_hevc_add_residual_16x16_10_neon
void ff_hevc_add_residual_16x16_10_neon(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride)
ff_hevc_add_residual_8x8_10_neon
void ff_hevc_add_residual_8x8_10_neon(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride)
ff_hevc_idct_8x8_dc_10_neon
void ff_hevc_idct_8x8_dc_10_neon(int16_t *coeffs)
ff_hevc_dsp_init_aarch64
av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth)
Definition: hevcdsp_init_aarch64.c:63
ff_hevc_add_residual_16x16_8_neon
void ff_hevc_add_residual_16x16_8_neon(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride)
ff_hevc_idct_16x16_10_neon
void ff_hevc_idct_16x16_10_neon(int16_t *coeffs, int col_limit)
ff_hevc_sao_band_filter_8x8_8_neon
void ff_hevc_sao_band_filter_8x8_8_neon(uint8_t *_dst, uint8_t *_src, ptrdiff_t stride_dst, ptrdiff_t stride_src, int16_t *sao_offset_val, int sao_left_class, int width, int height)
ff_hevc_add_residual_4x4_8_neon
void ff_hevc_add_residual_4x4_8_neon(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride)
cpu.h
ff_hevc_add_residual_32x32_10_neon
void ff_hevc_add_residual_32x32_10_neon(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride)