FFmpeg: libavcodec/aacpsdsp_template.c Source File

FFmpeg
aacpsdsp_template.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2010 Alex Converse <alex.converse@gmail.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  * Note: Rounding-to-nearest used unless otherwise stated
21  *
22  */
23 #include <stdint.h>
24 
25 #include "config.h"
26 #include "libavutil/attributes.h"
27 #include "aacpsdsp.h"
28 
29  static void ps_add_squares_c(INTFLOAT *restrict dst,
30  const INTFLOAT (*src)[2], int n)
31 {
32  for (int i = 0; i < n; i++)
33  dst[i] += (UINTFLOAT)AAC_MADD28(src[i][0], src[i][0], src[i][1], src[i][1]);
34 }
35 
36  static void ps_mul_pair_single_c(INTFLOAT (*restrict dst)[2],
37  INTFLOAT (*src0)[2], INTFLOAT *src1,
38  int n)
39 {
40  for (int i = 0; i < n; i++) {
41  dst[i][0] = AAC_MUL16(src0[i][0], src1[i]);
42  dst[i][1] = AAC_MUL16(src0[i][1], src1[i]);
43  }
44 }
45 
46  static void ps_hybrid_analysis_c(INTFLOAT (*restrict out)[2],
47  INTFLOAT (*in)[2],
48  const INTFLOAT (*filter)[8][2],
49  ptrdiff_t stride, int n)
50 {
51  INT64FLOAT inre0[6], inre1[6], inim0[6], inim1[6];
52 
53  for (int j = 0; j < 6; j++) {
54  inre0[j] = in[j][0] + in[12 - j][0];
55  inre1[j] = in[j][1] - in[12 - j][1];
56  inim0[j] = in[j][1] + in[12 - j][1];
57  inim1[j] = in[j][0] - in[12 - j][0];
58  }
59 
60  for (int i = 0; i < n; i++) {
61  INT64FLOAT sum_re = (INT64FLOAT)filter[i][6][0] * in[6][0];
62  INT64FLOAT sum_im = (INT64FLOAT)filter[i][6][0] * in[6][1];
63 
64  for (int j = 0; j < 6; j++) {
65  sum_re += (INT64FLOAT)filter[i][j][0] * inre0[j] -
66  (INT64FLOAT)filter[i][j][1] * inre1[j];
67  sum_im += (INT64FLOAT)filter[i][j][0] * inim0[j] +
68  (INT64FLOAT)filter[i][j][1] * inim1[j];
69  }
70 #if USE_FIXED
71  out[i * stride][0] = (int)((sum_re + 0x40000000) >> 31);
72  out[i * stride][1] = (int)((sum_im + 0x40000000) >> 31);
73 #else
74  out[i * stride][0] = sum_re;
75  out[i * stride][1] = sum_im;
76 #endif /* USE_FIXED */
77  }
78 }
79 
80  static void ps_hybrid_analysis_ileave_c(INTFLOAT (*restrict out)[32][2],
81  INTFLOAT L[2][38][64],
82  int i, int len)
83 {
84  for (; i < 64; i++) {
85  for (int j = 0; j < len; j++) {
86  out[i][j][0] = L[0][j][i];
87  out[i][j][1] = L[1][j][i];
88  }
89  }
90 }
91 
92  static void ps_hybrid_synthesis_deint_c(INTFLOAT out[2][38][64],
93  INTFLOAT (*restrict in)[32][2],
94  int i, int len)
95 {
96  for (; i < 64; i++) {
97  for (int n = 0; n < len; n++) {
98  out[0][n][i] = in[i][n][0];
99  out[1][n][i] = in[i][n][1];
100  }
101  }
102 }
103 
104  static void ps_decorrelate_c(INTFLOAT (*out)[2], INTFLOAT (*delay)[2],
105  INTFLOAT (*ap_delay)[PS_QMF_TIME_SLOTS + PS_MAX_AP_DELAY][2],
106  const INTFLOAT phi_fract[2], const INTFLOAT (*Q_fract)[2],
107  const INTFLOAT *transient_gain,
108  INTFLOAT g_decay_slope,
109  int len)
110 {
111  static const INTFLOAT a[] = { Q31(0.65143905753106f),
112  Q31(0.56471812200776f),
113  Q31(0.48954165955695f) };
114  INTFLOAT ag[PS_AP_LINKS];
115  int m, n;
116 
117  for (m = 0; m < PS_AP_LINKS; m++)
118  ag[m] = AAC_MUL30(a[m], g_decay_slope);
119 
120  for (n = 0; n < len; n++) {
121  INTFLOAT in_re = AAC_MSUB30(delay[n][0], phi_fract[0], delay[n][1], phi_fract[1]);
122  INTFLOAT in_im = AAC_MADD30(delay[n][0], phi_fract[1], delay[n][1], phi_fract[0]);
123  for (m = 0; m < PS_AP_LINKS; m++) {
124  INTFLOAT a_re = AAC_MUL31(ag[m], in_re);
125  INTFLOAT a_im = AAC_MUL31(ag[m], in_im);
126  INTFLOAT link_delay_re = ap_delay[m][n+2-m][0];
127  INTFLOAT link_delay_im = ap_delay[m][n+2-m][1];
128  INTFLOAT fractional_delay_re = Q_fract[m][0];
129  INTFLOAT fractional_delay_im = Q_fract[m][1];
130  INTFLOAT apd_re = in_re;
131  INTFLOAT apd_im = in_im;
132  in_re = AAC_MSUB30(link_delay_re, fractional_delay_re,
133  link_delay_im, fractional_delay_im);
134  in_re -= (UINTFLOAT)a_re;
135  in_im = AAC_MADD30(link_delay_re, fractional_delay_im,
136  link_delay_im, fractional_delay_re);
137  in_im -= (UINTFLOAT)a_im;
138  ap_delay[m][n+5][0] = apd_re + (UINTFLOAT)AAC_MUL31(ag[m], in_re);
139  ap_delay[m][n+5][1] = apd_im + (UINTFLOAT)AAC_MUL31(ag[m], in_im);
140  }
141  out[n][0] = AAC_MUL16(transient_gain[n], in_re);
142  out[n][1] = AAC_MUL16(transient_gain[n], in_im);
143  }
144 }
145 
146  static void ps_stereo_interpolate_c(INTFLOAT (*l)[2], INTFLOAT (*r)[2],
147  INTFLOAT h[2][4], INTFLOAT h_step[2][4],
148  int len)
149 {
150  INTFLOAT h0 = h[0][0];
151  INTFLOAT h1 = h[0][1];
152  INTFLOAT h2 = h[0][2];
153  INTFLOAT h3 = h[0][3];
154  UINTFLOAT hs0 = h_step[0][0];
155  UINTFLOAT hs1 = h_step[0][1];
156  UINTFLOAT hs2 = h_step[0][2];
157  UINTFLOAT hs3 = h_step[0][3];
158  int n;
159 
160  for (n = 0; n < len; n++) {
161  //l is s, r is d
162  INTFLOAT l_re = l[n][0];
163  INTFLOAT l_im = l[n][1];
164  INTFLOAT r_re = r[n][0];
165  INTFLOAT r_im = r[n][1];
166  h0 += hs0;
167  h1 += hs1;
168  h2 += hs2;
169  h3 += hs3;
170  l[n][0] = AAC_MADD30(h0, l_re, h2, r_re);
171  l[n][1] = AAC_MADD30(h0, l_im, h2, r_im);
172  r[n][0] = AAC_MADD30(h1, l_re, h3, r_re);
173  r[n][1] = AAC_MADD30(h1, l_im, h3, r_im);
174  }
175 }
176 
177  static void ps_stereo_interpolate_ipdopd_c(INTFLOAT (*l)[2], INTFLOAT (*r)[2],
178  INTFLOAT h[2][4], INTFLOAT h_step[2][4],
179  int len)
180 {
181  INTFLOAT h00 = h[0][0], h10 = h[1][0];
182  INTFLOAT h01 = h[0][1], h11 = h[1][1];
183  INTFLOAT h02 = h[0][2], h12 = h[1][2];
184  INTFLOAT h03 = h[0][3], h13 = h[1][3];
185  UINTFLOAT hs00 = h_step[0][0], hs10 = h_step[1][0];
186  UINTFLOAT hs01 = h_step[0][1], hs11 = h_step[1][1];
187  UINTFLOAT hs02 = h_step[0][2], hs12 = h_step[1][2];
188  UINTFLOAT hs03 = h_step[0][3], hs13 = h_step[1][3];
189  int n;
190 
191  for (n = 0; n < len; n++) {
192  //l is s, r is d
193  INTFLOAT l_re = l[n][0];
194  INTFLOAT l_im = l[n][1];
195  INTFLOAT r_re = r[n][0];
196  INTFLOAT r_im = r[n][1];
197  h00 += hs00;
198  h01 += hs01;
199  h02 += hs02;
200  h03 += hs03;
201  h10 += hs10;
202  h11 += hs11;
203  h12 += hs12;
204  h13 += hs13;
205 
206  l[n][0] = AAC_MSUB30_V8(h00, l_re, h02, r_re, h10, l_im, h12, r_im);
207  l[n][1] = AAC_MADD30_V8(h00, l_im, h02, r_im, h10, l_re, h12, r_re);
208  r[n][0] = AAC_MSUB30_V8(h01, l_re, h03, r_re, h11, l_im, h13, r_im);
209  r[n][1] = AAC_MADD30_V8(h01, l_im, h03, r_im, h11, l_re, h13, r_re);
210  }
211 }
212 
213  av_cold void AAC_RENAME(ff_psdsp_init)(PSDSPContext *s)
214 {
215  s->add_squares = ps_add_squares_c;
216  s->mul_pair_single = ps_mul_pair_single_c;
217  s->hybrid_analysis = ps_hybrid_analysis_c;
218  s->hybrid_analysis_ileave = ps_hybrid_analysis_ileave_c;
219  s->hybrid_synthesis_deint = ps_hybrid_synthesis_deint_c;
220  s->decorrelate = ps_decorrelate_c;
221  s->stereo_interpolate[0] = ps_stereo_interpolate_c;
222  s->stereo_interpolate[1] = ps_stereo_interpolate_ipdopd_c;
223 
224 #if !USE_FIXED
225 #if ARCH_ARM
226  ff_psdsp_init_arm(s);
227 #elif ARCH_AARCH64
228  ff_psdsp_init_aarch64(s);
229 #elif ARCH_RISCV
230  ff_psdsp_init_riscv(s);
231 #elif ARCH_X86
232  ff_psdsp_init_x86(s);
233 #endif
234 #endif /* !USE_FIXED */
235 }
AAC_MADD30
#define AAC_MADD30(x, y, a, b)
Definition: aac_defines.h:118
h0
static const float h0[64]
Definition: speexdata.h:741
r
const char * r
Definition: vf_curves.c:127
ps_hybrid_analysis_c
static void ps_hybrid_analysis_c(INTFLOAT(*restrict out)[2], INTFLOAT(*in)[2], const INTFLOAT(*filter)[8][2], ptrdiff_t stride, int n)
Definition: aacpsdsp_template.c:46
out
FILE * out
Definition: movenc.c:55
ps_add_squares_c
static void ps_add_squares_c(INTFLOAT *restrict dst, const INTFLOAT(*src)[2], int n)
Definition: aacpsdsp_template.c:29
src1
const pixel * src1
Definition: h264pred_template.c:420
ff_psdsp_init_x86
void ff_psdsp_init_x86(PSDSPContext *s)
Definition: aacpsdsp_init.c:52
ps_hybrid_analysis_ileave_c
static void ps_hybrid_analysis_ileave_c(INTFLOAT(*restrict out)[32][2], INTFLOAT L[2][38][64], int i, int len)
Definition: aacpsdsp_template.c:80
filter
void(* filter)(uint8_t *src, int stride, int qscale)
Definition: h263dsp.c:29
PS_QMF_TIME_SLOTS
#define PS_QMF_TIME_SLOTS
Definition: aacps.h:37
ps_stereo_interpolate_c
static void ps_stereo_interpolate_c(INTFLOAT(*l)[2], INTFLOAT(*r)[2], INTFLOAT h[2][4], INTFLOAT h_step[2][4], int len)
Definition: aacpsdsp_template.c:146
ff_psdsp_init_arm
void ff_psdsp_init_arm(PSDSPContext *s)
Definition: aacpsdsp_init_arm.c:46
ps_mul_pair_single_c
static void ps_mul_pair_single_c(INTFLOAT(*restrict dst)[2], INTFLOAT(*src0)[2], INTFLOAT *src1, int n)
Definition: aacpsdsp_template.c:36
av_cold
#define av_cold
Definition: attributes.h:106
aacpsdsp.h
AAC_MUL31
#define AAC_MUL31(x, y)
Definition: aac_defines.h:116
ff_psdsp_init_aarch64
void ff_psdsp_init_aarch64(PSDSPContext *s)
Definition: aacpsdsp_init_aarch64.c:38
s
#define s(width, name)
Definition: cbs_vp9.c:198
PS_AP_LINKS
#define PS_AP_LINKS
Definition: aacps.h:39
AAC_MUL30
#define AAC_MUL30(x, y)
Definition: aac_defines.h:115
AAC_MSUB30
#define AAC_MSUB30(x, y, a, b)
Definition: aac_defines.h:121
ps_stereo_interpolate_ipdopd_c
static void ps_stereo_interpolate_ipdopd_c(INTFLOAT(*l)[2], INTFLOAT(*r)[2], INTFLOAT h[2][4], INTFLOAT h_step[2][4], int len)
Definition: aacpsdsp_template.c:177
PS_MAX_AP_DELAY
#define PS_MAX_AP_DELAY
Definition: aacps.h:40
f
f
Definition: af_crystalizer.c:122
dst
uint8_t ptrdiff_t const uint8_t ptrdiff_t int intptr_t intptr_t int int16_t * dst
Definition: dsp.h:87
ps_decorrelate_c
static void ps_decorrelate_c(INTFLOAT(*out)[2], INTFLOAT(*delay)[2], INTFLOAT(*ap_delay)[PS_QMF_TIME_SLOTS+PS_MAX_AP_DELAY][2], const INTFLOAT phi_fract[2], const INTFLOAT(*Q_fract)[2], const INTFLOAT *transient_gain, INTFLOAT g_decay_slope, int len)
Definition: aacpsdsp_template.c:104
a
The reader does not expect b to be semantically here and if the code is changed by maybe adding a a division or other the signedness will almost certainly be mistaken To avoid this confusion a new type was SUINT is the C unsigned type but it holds a signed int to use the same example SUINT a
Definition: undefined.txt:41
INT64FLOAT
float INT64FLOAT
Definition: aac_defines.h:103
attributes.h
i
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:256
AAC_MADD30_V8
#define AAC_MADD30_V8(x, y, a, b, c, d, e, f)
Definition: aac_defines.h:119
AAC_MUL16
#define AAC_MUL16(x, y)
Definition: aac_defines.h:113
AAC_RENAME
#define AAC_RENAME(x)
Definition: aac_defines.h:99
len
int len
Definition: vorbis_enc_data.h:426
UINTFLOAT
float UINTFLOAT
Definition: aac_defines.h:102
stride
#define stride
Definition: h264pred_template.c:536
ff_psdsp_init
av_cold void AAC_RENAME() ff_psdsp_init(PSDSPContext *s)
Definition: aacpsdsp_template.c:213
phi_fract
static int phi_fract[2][50][2]
Definition: aacps_fixed_tablegen.h:61
PSDSPContext
Definition: aacpsdsp.h:32
L
#define L(x)
Definition: vpx_arith.h:36
ff_psdsp_init_riscv
void ff_psdsp_init_riscv(PSDSPContext *s)
Definition: aacpsdsp_init.c:40
Q31
#define Q31(x)
Definition: aac_defines.h:111
src0
const pixel *const src0
Definition: h264pred_template.c:419
AAC_MSUB30_V8
#define AAC_MSUB30_V8(x, y, a, b, c, d, e, f)
Definition: aac_defines.h:122
h
h
Definition: vp9dsp_template.c:2070
AAC_MADD28
#define AAC_MADD28(x, y, a, b)
Definition: aac_defines.h:117
INTFLOAT
float INTFLOAT
Definition: aac_defines.h:101
ps_hybrid_synthesis_deint_c
static void ps_hybrid_synthesis_deint_c(INTFLOAT out[2][38][64], INTFLOAT(*restrict in)[32][2], int i, int len)
Definition: aacpsdsp_template.c:92
src
#define src
Definition: vp8dsp.c:248
Generated on Sat Oct 18 2025 19:21:47 for FFmpeg by   doxygen 1.8.17

AltStyle によって変換されたページ (->オリジナル) /