FFmpeg
hevcdsp_init_aarch64.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2020 Reimar Döffinger
3  * Copyright (c) 2023 xu fulong <839789740@qq.com>
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 #include <stdint.h>
23 
24 #include "libavutil/attributes.h"
25 #include "libavutil/cpu.h"
26 #include "libavutil/aarch64/cpu.h"
27 #include "libavcodec/hevcdsp.h"
28 
29 void ff_hevc_v_loop_filter_chroma_8_neon(uint8_t *_pix, ptrdiff_t _stride,
30  const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q);
31 void ff_hevc_v_loop_filter_chroma_10_neon(uint8_t *_pix, ptrdiff_t _stride,
32  const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q);
33 void ff_hevc_v_loop_filter_chroma_12_neon(uint8_t *_pix, ptrdiff_t _stride,
34  const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q);
35 void ff_hevc_h_loop_filter_chroma_8_neon(uint8_t *_pix, ptrdiff_t _stride,
36  const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q);
37 void ff_hevc_h_loop_filter_chroma_10_neon(uint8_t *_pix, ptrdiff_t _stride,
38  const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q);
39 void ff_hevc_h_loop_filter_chroma_12_neon(uint8_t *_pix, ptrdiff_t _stride,
40  const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q);
41 void ff_hevc_add_residual_4x4_8_neon(uint8_t *_dst, const int16_t *coeffs,
42  ptrdiff_t stride);
43 void ff_hevc_add_residual_4x4_10_neon(uint8_t *_dst, const int16_t *coeffs,
44  ptrdiff_t stride);
45 void ff_hevc_add_residual_4x4_12_neon(uint8_t *_dst, const int16_t *coeffs,
46  ptrdiff_t stride);
47 void ff_hevc_add_residual_8x8_8_neon(uint8_t *_dst, const int16_t *coeffs,
48  ptrdiff_t stride);
49 void ff_hevc_add_residual_8x8_10_neon(uint8_t *_dst, const int16_t *coeffs,
50  ptrdiff_t stride);
51 void ff_hevc_add_residual_8x8_12_neon(uint8_t *_dst, const int16_t *coeffs,
52  ptrdiff_t stride);
53 void ff_hevc_add_residual_16x16_8_neon(uint8_t *_dst, const int16_t *coeffs,
54  ptrdiff_t stride);
55 void ff_hevc_add_residual_16x16_10_neon(uint8_t *_dst, const int16_t *coeffs,
56  ptrdiff_t stride);
57 void ff_hevc_add_residual_16x16_12_neon(uint8_t *_dst, const int16_t *coeffs,
58  ptrdiff_t stride);
59 void ff_hevc_add_residual_32x32_8_neon(uint8_t *_dst, const int16_t *coeffs,
60  ptrdiff_t stride);
61 void ff_hevc_add_residual_32x32_10_neon(uint8_t *_dst, const int16_t *coeffs,
62  ptrdiff_t stride);
63 void ff_hevc_add_residual_32x32_12_neon(uint8_t *_dst, const int16_t *coeffs,
64  ptrdiff_t stride);
65 void ff_hevc_idct_4x4_8_neon(int16_t *coeffs, int col_limit);
66 void ff_hevc_idct_4x4_10_neon(int16_t *coeffs, int col_limit);
67 void ff_hevc_idct_8x8_8_neon(int16_t *coeffs, int col_limit);
68 void ff_hevc_idct_8x8_10_neon(int16_t *coeffs, int col_limit);
69 void ff_hevc_idct_16x16_8_neon(int16_t *coeffs, int col_limit);
70 void ff_hevc_idct_16x16_10_neon(int16_t *coeffs, int col_limit);
71 void ff_hevc_idct_32x32_8_neon(int16_t *coeffs, int col_limit);
72 void ff_hevc_idct_32x32_10_neon(int16_t *coeffs, int col_limit);
73 void ff_hevc_idct_4x4_dc_8_neon(int16_t *coeffs);
74 void ff_hevc_idct_8x8_dc_8_neon(int16_t *coeffs);
75 void ff_hevc_idct_16x16_dc_8_neon(int16_t *coeffs);
76 void ff_hevc_idct_32x32_dc_8_neon(int16_t *coeffs);
77 void ff_hevc_idct_4x4_dc_10_neon(int16_t *coeffs);
78 void ff_hevc_idct_8x8_dc_10_neon(int16_t *coeffs);
79 void ff_hevc_idct_16x16_dc_10_neon(int16_t *coeffs);
80 void ff_hevc_idct_32x32_dc_10_neon(int16_t *coeffs);
81 void ff_hevc_transform_luma_4x4_neon_8(int16_t *coeffs);
82 void ff_hevc_sao_band_filter_8x8_8_neon(uint8_t *_dst, const uint8_t *_src,
83  ptrdiff_t stride_dst, ptrdiff_t stride_src,
84  const int16_t *sao_offset_val, int sao_left_class,
85  int width, int height);
86 void ff_hevc_sao_edge_filter_16x16_8_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride_dst,
87  const int16_t *sao_offset_val, int eo, int width, int height);
88 void ff_hevc_sao_edge_filter_8x8_8_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride_dst,
89  const int16_t *sao_offset_val, int eo, int width, int height);
90 void ff_hevc_put_hevc_qpel_h4_8_neon(int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height,
91  intptr_t mx, intptr_t my, int width);
92 void ff_hevc_put_hevc_qpel_h6_8_neon(int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height,
93  intptr_t mx, intptr_t my, int width);
94 void ff_hevc_put_hevc_qpel_h8_8_neon(int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height,
95  intptr_t mx, intptr_t my, int width);
96 void ff_hevc_put_hevc_qpel_h12_8_neon(int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height,
97  intptr_t mx, intptr_t my, int width);
98 void ff_hevc_put_hevc_qpel_h16_8_neon(int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height,
99  intptr_t mx, intptr_t my, int width);
100 void ff_hevc_put_hevc_qpel_uni_h4_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src,
101  ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my,
102  int width);
103 void ff_hevc_put_hevc_qpel_uni_h6_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src,
104  ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my,
105  int width);
106 void ff_hevc_put_hevc_qpel_uni_h8_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src,
107  ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my,
108  int width);
109 void ff_hevc_put_hevc_qpel_uni_h12_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src,
110  ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t
111  my, int width);
112 void ff_hevc_put_hevc_qpel_uni_h16_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src,
113  ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t
114  my, int width);
115 void ff_hevc_put_hevc_qpel_bi_h4_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src,
116  ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t
117  mx, intptr_t my, int width);
118 void ff_hevc_put_hevc_qpel_bi_h6_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src,
119  ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t
120  mx, intptr_t my, int width);
121 void ff_hevc_put_hevc_qpel_bi_h8_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src,
122  ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t
123  mx, intptr_t my, int width);
124 void ff_hevc_put_hevc_qpel_bi_h12_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src,
125  ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t
126  mx, intptr_t my, int width);
127 void ff_hevc_put_hevc_qpel_bi_h16_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src,
128  ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t
129  mx, intptr_t my, int width);
130 
131 #define NEON8_FNPROTO(fn, args, ext) \
132  void ff_hevc_put_hevc_##fn##4_8_neon##ext args; \
133  void ff_hevc_put_hevc_##fn##6_8_neon##ext args; \
134  void ff_hevc_put_hevc_##fn##8_8_neon##ext args; \
135  void ff_hevc_put_hevc_##fn##12_8_neon##ext args; \
136  void ff_hevc_put_hevc_##fn##16_8_neon##ext args; \
137  void ff_hevc_put_hevc_##fn##24_8_neon##ext args; \
138  void ff_hevc_put_hevc_##fn##32_8_neon##ext args; \
139  void ff_hevc_put_hevc_##fn##48_8_neon##ext args; \
140  void ff_hevc_put_hevc_##fn##64_8_neon##ext args
141 
142 #define NEON8_FNPROTO_PARTIAL_4(fn, args, ext) \
143  void ff_hevc_put_hevc_##fn##4_8_neon##ext args; \
144  void ff_hevc_put_hevc_##fn##8_8_neon##ext args; \
145  void ff_hevc_put_hevc_##fn##16_8_neon##ext args; \
146  void ff_hevc_put_hevc_##fn##64_8_neon##ext args
147 
148 #define NEON8_FNPROTO_PARTIAL_5(fn, args, ext) \
149  void ff_hevc_put_hevc_##fn##4_8_neon##ext args; \
150  void ff_hevc_put_hevc_##fn##8_8_neon##ext args; \
151  void ff_hevc_put_hevc_##fn##16_8_neon##ext args; \
152  void ff_hevc_put_hevc_##fn##32_8_neon##ext args; \
153  void ff_hevc_put_hevc_##fn##64_8_neon##ext args
154 
155 NEON8_FNPROTO(pel_uni_pixels, (uint8_t *_dst, ptrdiff_t _dststride,
156  const uint8_t *_src, ptrdiff_t _srcstride,
157  int height, intptr_t mx, intptr_t my, int width),);
158 
159 NEON8_FNPROTO(pel_uni_w_pixels, (uint8_t *_dst, ptrdiff_t _dststride,
160  const uint8_t *_src, ptrdiff_t _srcstride,
161  int height, int denom, int wx, int ox,
162  intptr_t mx, intptr_t my, int width),);
163 
164 NEON8_FNPROTO(epel_uni_v, (uint8_t *dst, ptrdiff_t dststride,
165  const uint8_t *src, ptrdiff_t srcstride,
166  int height, intptr_t mx, intptr_t my, int width),);
167 
168 NEON8_FNPROTO(epel_uni_hv, (uint8_t *dst, ptrdiff_t _dststride,
169  const uint8_t *src, ptrdiff_t srcstride,
170  int height, intptr_t mx, intptr_t my, int width), _i8mm);
171 
172 NEON8_FNPROTO(epel_uni_w_v, (uint8_t *_dst, ptrdiff_t _dststride,
173  const uint8_t *_src, ptrdiff_t _srcstride,
174  int height, int denom, int wx, int ox,
175  intptr_t mx, intptr_t my, int width),);
176 
177 NEON8_FNPROTO_PARTIAL_4(qpel_uni_w_v, (uint8_t *_dst, ptrdiff_t _dststride,
178  const uint8_t *_src, ptrdiff_t _srcstride,
179  int height, int denom, int wx, int ox,
180  intptr_t mx, intptr_t my, int width),);
181 
182 NEON8_FNPROTO(epel_h, (int16_t *dst,
183  const uint8_t *_src, ptrdiff_t _srcstride,
184  int height, intptr_t mx, intptr_t my, int width), _i8mm);
185 
186 NEON8_FNPROTO(epel_uni_w_h, (uint8_t *_dst, ptrdiff_t _dststride,
187  const uint8_t *_src, ptrdiff_t _srcstride,
188  int height, int denom, int wx, int ox,
189  intptr_t mx, intptr_t my, int width), _i8mm);
190 
191 NEON8_FNPROTO(qpel_h, (int16_t *dst,
192  const uint8_t *_src, ptrdiff_t _srcstride,
193  int height, intptr_t mx, intptr_t my, int width), _i8mm);
194 
195 NEON8_FNPROTO(qpel_uni_v, (uint8_t *dst, ptrdiff_t dststride,
196  const uint8_t *src, ptrdiff_t srcstride,
197  int height, intptr_t mx, intptr_t my, int width),);
198 
199 NEON8_FNPROTO(qpel_uni_hv, (uint8_t *dst, ptrdiff_t dststride,
200  const uint8_t *src, ptrdiff_t srcstride,
201  int height, intptr_t mx, intptr_t my, int width), _i8mm);
202 
203 NEON8_FNPROTO(qpel_uni_w_h, (uint8_t *_dst, ptrdiff_t _dststride,
204  const uint8_t *_src, ptrdiff_t _srcstride,
205  int height, int denom, int wx, int ox,
206  intptr_t mx, intptr_t my, int width), _i8mm);
207 
208 NEON8_FNPROTO(epel_uni_w_hv, (uint8_t *_dst, ptrdiff_t _dststride,
209  const uint8_t *_src, ptrdiff_t _srcstride,
210  int height, int denom, int wx, int ox,
211  intptr_t mx, intptr_t my, int width), _i8mm);
212 
213 NEON8_FNPROTO_PARTIAL_5(qpel_uni_w_hv, (uint8_t *_dst, ptrdiff_t _dststride,
214  const uint8_t *_src, ptrdiff_t _srcstride,
215  int height, int denom, int wx, int ox,
216  intptr_t mx, intptr_t my, int width), _i8mm);
217 
218 #define NEON8_FNASSIGN(member, v, h, fn, ext) \
219  member[1][v][h] = ff_hevc_put_hevc_##fn##4_8_neon##ext; \
220  member[2][v][h] = ff_hevc_put_hevc_##fn##6_8_neon##ext; \
221  member[3][v][h] = ff_hevc_put_hevc_##fn##8_8_neon##ext; \
222  member[4][v][h] = ff_hevc_put_hevc_##fn##12_8_neon##ext; \
223  member[5][v][h] = ff_hevc_put_hevc_##fn##16_8_neon##ext; \
224  member[6][v][h] = ff_hevc_put_hevc_##fn##24_8_neon##ext; \
225  member[7][v][h] = ff_hevc_put_hevc_##fn##32_8_neon##ext; \
226  member[8][v][h] = ff_hevc_put_hevc_##fn##48_8_neon##ext; \
227  member[9][v][h] = ff_hevc_put_hevc_##fn##64_8_neon##ext;
228 
229 #define NEON8_FNASSIGN_PARTIAL_4(member, v, h, fn, ext) \
230  member[1][v][h] = ff_hevc_put_hevc_##fn##4_8_neon##ext; \
231  member[3][v][h] = ff_hevc_put_hevc_##fn##8_8_neon##ext; \
232  member[5][v][h] = ff_hevc_put_hevc_##fn##16_8_neon##ext; \
233  member[7][v][h] = ff_hevc_put_hevc_##fn##64_8_neon##ext; \
234  member[8][v][h] = ff_hevc_put_hevc_##fn##64_8_neon##ext; \
235  member[9][v][h] = ff_hevc_put_hevc_##fn##64_8_neon##ext;
236 
237 #define NEON8_FNASSIGN_PARTIAL_5(member, v, h, fn, ext) \
238  member[1][v][h] = ff_hevc_put_hevc_##fn##4_8_neon##ext; \
239  member[3][v][h] = ff_hevc_put_hevc_##fn##8_8_neon##ext; \
240  member[5][v][h] = ff_hevc_put_hevc_##fn##16_8_neon##ext; \
241  member[7][v][h] = ff_hevc_put_hevc_##fn##32_8_neon##ext; \
242  member[9][v][h] = ff_hevc_put_hevc_##fn##64_8_neon##ext;
243 
244 av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth)
245 {
246  int cpu_flags = av_get_cpu_flags();
247  if (!have_neon(cpu_flags)) return;
248 
249  if (bit_depth == 8) {
250  c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_8_neon;
251  c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_8_neon;
252  c->add_residual[0] = ff_hevc_add_residual_4x4_8_neon;
253  c->add_residual[1] = ff_hevc_add_residual_8x8_8_neon;
254  c->add_residual[2] = ff_hevc_add_residual_16x16_8_neon;
255  c->add_residual[3] = ff_hevc_add_residual_32x32_8_neon;
256  c->idct[0] = ff_hevc_idct_4x4_8_neon;
257  c->idct[1] = ff_hevc_idct_8x8_8_neon;
258  c->idct[2] = ff_hevc_idct_16x16_8_neon;
259  c->idct[3] = ff_hevc_idct_32x32_8_neon;
260  c->idct_dc[0] = ff_hevc_idct_4x4_dc_8_neon;
261  c->idct_dc[1] = ff_hevc_idct_8x8_dc_8_neon;
262  c->idct_dc[2] = ff_hevc_idct_16x16_dc_8_neon;
263  c->idct_dc[3] = ff_hevc_idct_32x32_dc_8_neon;
264  c->transform_4x4_luma = ff_hevc_transform_luma_4x4_neon_8;
265  c->sao_band_filter[0] =
266  c->sao_band_filter[1] =
267  c->sao_band_filter[2] =
268  c->sao_band_filter[3] =
269  c->sao_band_filter[4] = ff_hevc_sao_band_filter_8x8_8_neon;
270  c->sao_edge_filter[0] = ff_hevc_sao_edge_filter_8x8_8_neon;
271  c->sao_edge_filter[1] =
272  c->sao_edge_filter[2] =
273  c->sao_edge_filter[3] =
274  c->sao_edge_filter[4] = ff_hevc_sao_edge_filter_16x16_8_neon;
275  c->put_hevc_qpel[1][0][1] = ff_hevc_put_hevc_qpel_h4_8_neon;
276  c->put_hevc_qpel[2][0][1] = ff_hevc_put_hevc_qpel_h6_8_neon;
277  c->put_hevc_qpel[3][0][1] = ff_hevc_put_hevc_qpel_h8_8_neon;
278  c->put_hevc_qpel[4][0][1] =
279  c->put_hevc_qpel[6][0][1] = ff_hevc_put_hevc_qpel_h12_8_neon;
280  c->put_hevc_qpel[5][0][1] =
281  c->put_hevc_qpel[7][0][1] =
282  c->put_hevc_qpel[8][0][1] =
283  c->put_hevc_qpel[9][0][1] = ff_hevc_put_hevc_qpel_h16_8_neon;
284  c->put_hevc_qpel_uni[1][0][1] = ff_hevc_put_hevc_qpel_uni_h4_8_neon;
285  c->put_hevc_qpel_uni[2][0][1] = ff_hevc_put_hevc_qpel_uni_h6_8_neon;
286  c->put_hevc_qpel_uni[3][0][1] = ff_hevc_put_hevc_qpel_uni_h8_8_neon;
287  c->put_hevc_qpel_uni[4][0][1] =
288  c->put_hevc_qpel_uni[6][0][1] = ff_hevc_put_hevc_qpel_uni_h12_8_neon;
289  c->put_hevc_qpel_uni[5][0][1] =
290  c->put_hevc_qpel_uni[7][0][1] =
291  c->put_hevc_qpel_uni[8][0][1] =
292  c->put_hevc_qpel_uni[9][0][1] = ff_hevc_put_hevc_qpel_uni_h16_8_neon;
293  c->put_hevc_qpel_bi[1][0][1] = ff_hevc_put_hevc_qpel_bi_h4_8_neon;
294  c->put_hevc_qpel_bi[2][0][1] = ff_hevc_put_hevc_qpel_bi_h6_8_neon;
295  c->put_hevc_qpel_bi[3][0][1] = ff_hevc_put_hevc_qpel_bi_h8_8_neon;
296  c->put_hevc_qpel_bi[4][0][1] =
297  c->put_hevc_qpel_bi[6][0][1] = ff_hevc_put_hevc_qpel_bi_h12_8_neon;
298  c->put_hevc_qpel_bi[5][0][1] =
299  c->put_hevc_qpel_bi[7][0][1] =
300  c->put_hevc_qpel_bi[8][0][1] =
301  c->put_hevc_qpel_bi[9][0][1] = ff_hevc_put_hevc_qpel_bi_h16_8_neon;
302 
303  NEON8_FNASSIGN(c->put_hevc_epel_uni, 0, 0, pel_uni_pixels,);
304  NEON8_FNASSIGN(c->put_hevc_epel_uni, 1, 0, epel_uni_v,);
305  NEON8_FNASSIGN(c->put_hevc_qpel_uni, 0, 0, pel_uni_pixels,);
306  NEON8_FNASSIGN(c->put_hevc_qpel_uni, 1, 0, qpel_uni_v,);
307  NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 0, 0, pel_uni_w_pixels,);
308  NEON8_FNASSIGN(c->put_hevc_qpel_uni_w, 0, 0, pel_uni_w_pixels,);
309  NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 1, 0, epel_uni_w_v,);
310  NEON8_FNASSIGN_PARTIAL_4(c->put_hevc_qpel_uni_w, 1, 0, qpel_uni_w_v,);
311 
312  if (have_i8mm(cpu_flags)) {
313  NEON8_FNASSIGN(c->put_hevc_epel, 0, 1, epel_h, _i8mm);
314  NEON8_FNASSIGN(c->put_hevc_epel_uni, 1, 1, epel_uni_hv, _i8mm);
315  NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 0, 1, epel_uni_w_h ,_i8mm);
316  NEON8_FNASSIGN(c->put_hevc_qpel, 0, 1, qpel_h, _i8mm);
317  NEON8_FNASSIGN(c->put_hevc_qpel_uni, 1, 1, qpel_uni_hv, _i8mm);
318  NEON8_FNASSIGN(c->put_hevc_qpel_uni_w, 0, 1, qpel_uni_w_h, _i8mm);
319  NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 1, 1, epel_uni_w_hv, _i8mm);
320  NEON8_FNASSIGN_PARTIAL_5(c->put_hevc_qpel_uni_w, 1, 1, qpel_uni_w_hv, _i8mm);
321  }
322 
323  }
324  if (bit_depth == 10) {
325  c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_10_neon;
326  c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_10_neon;
327  c->add_residual[0] = ff_hevc_add_residual_4x4_10_neon;
328  c->add_residual[1] = ff_hevc_add_residual_8x8_10_neon;
329  c->add_residual[2] = ff_hevc_add_residual_16x16_10_neon;
330  c->add_residual[3] = ff_hevc_add_residual_32x32_10_neon;
331  c->idct[0] = ff_hevc_idct_4x4_10_neon;
332  c->idct[1] = ff_hevc_idct_8x8_10_neon;
333  c->idct[2] = ff_hevc_idct_16x16_10_neon;
334  c->idct[3] = ff_hevc_idct_32x32_10_neon;
335  c->idct_dc[0] = ff_hevc_idct_4x4_dc_10_neon;
336  c->idct_dc[1] = ff_hevc_idct_8x8_dc_10_neon;
337  c->idct_dc[2] = ff_hevc_idct_16x16_dc_10_neon;
338  c->idct_dc[3] = ff_hevc_idct_32x32_dc_10_neon;
339  }
340  if (bit_depth == 12) {
341  c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_12_neon;
342  c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_12_neon;
343  c->add_residual[0] = ff_hevc_add_residual_4x4_12_neon;
344  c->add_residual[1] = ff_hevc_add_residual_8x8_12_neon;
345  c->add_residual[2] = ff_hevc_add_residual_16x16_12_neon;
346  c->add_residual[3] = ff_hevc_add_residual_32x32_12_neon;
347  }
348 }
NEON8_FNPROTO_PARTIAL_5
#define NEON8_FNPROTO_PARTIAL_5(fn, args, ext)
Definition: hevcdsp_init_aarch64.c:148
ff_hevc_v_loop_filter_chroma_10_neon
void ff_hevc_v_loop_filter_chroma_10_neon(uint8_t *_pix, ptrdiff_t _stride, const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q)
NEON8_FNASSIGN
#define NEON8_FNASSIGN(member, v, h, fn, ext)
Definition: hevcdsp_init_aarch64.c:218
ff_hevc_idct_4x4_dc_8_neon
void ff_hevc_idct_4x4_dc_8_neon(int16_t *coeffs)
ff_hevc_put_hevc_qpel_h4_8_neon
void ff_hevc_put_hevc_qpel_h4_8_neon(int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width)
ff_hevc_put_hevc_qpel_bi_h6_8_neon
void ff_hevc_put_hevc_qpel_bi_h6_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width)
ff_hevc_transform_luma_4x4_neon_8
void ff_hevc_transform_luma_4x4_neon_8(int16_t *coeffs)
ff_hevc_sao_edge_filter_8x8_8_neon
void ff_hevc_sao_edge_filter_8x8_8_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride_dst, const int16_t *sao_offset_val, int eo, int width, int height)
_stride
ptrdiff_t _stride
Definition: h264pred_template.c:411
ff_hevc_idct_16x16_8_neon
void ff_hevc_idct_16x16_8_neon(int16_t *coeffs, int col_limit)
ff_hevc_put_hevc_qpel_bi_h12_8_neon
void ff_hevc_put_hevc_qpel_bi_h12_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width)
ff_hevc_put_hevc_qpel_h8_8_neon
void ff_hevc_put_hevc_qpel_h8_8_neon(int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width)
av_get_cpu_flags
int av_get_cpu_flags(void)
Return the flags which specify extensions supported by the CPU.
Definition: cpu.c:103
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_hevc_add_residual_16x16_10_neon
void ff_hevc_add_residual_16x16_10_neon(uint8_t *_dst, const int16_t *coeffs, ptrdiff_t stride)
ff_hevc_idct_32x32_8_neon
void ff_hevc_idct_32x32_8_neon(int16_t *coeffs, int col_limit)
ff_hevc_add_residual_4x4_10_neon
void ff_hevc_add_residual_4x4_10_neon(uint8_t *_dst, const int16_t *coeffs, ptrdiff_t stride)
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
ff_hevc_idct_32x32_10_neon
void ff_hevc_idct_32x32_10_neon(int16_t *coeffs, int col_limit)
ff_hevc_add_residual_32x32_12_neon
void ff_hevc_add_residual_32x32_12_neon(uint8_t *_dst, const int16_t *coeffs, ptrdiff_t stride)
ff_hevc_put_hevc_qpel_uni_h8_8_neon
void ff_hevc_put_hevc_qpel_uni_h8_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width)
ff_hevc_put_hevc_qpel_uni_h12_8_neon
void ff_hevc_put_hevc_qpel_uni_h12_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width)
ff_hevc_put_hevc_qpel_bi_h8_8_neon
void ff_hevc_put_hevc_qpel_bi_h8_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width)
hevcdsp.h
NEON8_FNASSIGN_PARTIAL_5
#define NEON8_FNASSIGN_PARTIAL_5(member, v, h, fn, ext)
Definition: hevcdsp_init_aarch64.c:237
ff_hevc_put_hevc_qpel_bi_h4_8_neon
void ff_hevc_put_hevc_qpel_bi_h4_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width)
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)
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
have_i8mm
#define have_i8mm(flags)
Definition: cpu.h:29
ff_hevc_put_hevc_qpel_uni_h4_8_neon
void ff_hevc_put_hevc_qpel_uni_h4_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width)
ff_hevc_sao_band_filter_8x8_8_neon
void ff_hevc_sao_band_filter_8x8_8_neon(uint8_t *_dst, const uint8_t *_src, ptrdiff_t stride_dst, ptrdiff_t stride_src, const int16_t *sao_offset_val, int sao_left_class, int width, int height)
ff_hevc_add_residual_8x8_8_neon
void ff_hevc_add_residual_8x8_8_neon(uint8_t *_dst, const int16_t *coeffs, ptrdiff_t stride)
ff_hevc_v_loop_filter_chroma_12_neon
void ff_hevc_v_loop_filter_chroma_12_neon(uint8_t *_pix, ptrdiff_t _stride, const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q)
cpu.h
ff_hevc_add_residual_16x16_12_neon
void ff_hevc_add_residual_16x16_12_neon(uint8_t *_dst, const int16_t *coeffs, ptrdiff_t stride)
ff_hevc_idct_4x4_10_neon
void ff_hevc_idct_4x4_10_neon(int16_t *coeffs, int col_limit)
ff_hevc_add_residual_8x8_12_neon
void ff_hevc_add_residual_8x8_12_neon(uint8_t *_dst, const int16_t *coeffs, ptrdiff_t stride)
ff_hevc_put_hevc_qpel_h6_8_neon
void ff_hevc_put_hevc_qpel_h6_8_neon(int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width)
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
ff_hevc_h_loop_filter_chroma_12_neon
void ff_hevc_h_loop_filter_chroma_12_neon(uint8_t *_pix, ptrdiff_t _stride, const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q)
NEON8_FNPROTO
#define NEON8_FNPROTO(fn, args, ext)
Definition: hevcdsp_init_aarch64.c:131
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_4x4_8_neon
void ff_hevc_idct_4x4_8_neon(int16_t *coeffs, int col_limit)
ff_hevc_idct_32x32_dc_8_neon
void ff_hevc_idct_32x32_dc_8_neon(int16_t *coeffs)
src2
const pixel * src2
Definition: h264pred_template.c:422
ff_hevc_idct_8x8_dc_8_neon
void ff_hevc_idct_8x8_dc_8_neon(int16_t *coeffs)
ff_hevc_put_hevc_qpel_uni_h6_8_neon
void ff_hevc_put_hevc_qpel_uni_h6_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width)
ff_hevc_put_hevc_qpel_h12_8_neon
void ff_hevc_put_hevc_qpel_h12_8_neon(int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width)
ff_hevc_put_hevc_qpel_uni_h16_8_neon
void ff_hevc_put_hevc_qpel_uni_h16_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width)
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:244
stride
#define stride
Definition: h264pred_template.c:537
ff_hevc_sao_edge_filter_16x16_8_neon
void ff_hevc_sao_edge_filter_16x16_8_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride_dst, const int16_t *sao_offset_val, int eo, int width, int height)
ff_hevc_add_residual_32x32_8_neon
void ff_hevc_add_residual_32x32_8_neon(uint8_t *_dst, const int16_t *coeffs, ptrdiff_t stride)
ff_hevc_v_loop_filter_chroma_8_neon
void ff_hevc_v_loop_filter_chroma_8_neon(uint8_t *_pix, ptrdiff_t _stride, const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q)
ff_hevc_put_hevc_qpel_h16_8_neon
void ff_hevc_put_hevc_qpel_h16_8_neon(int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width)
ff_hevc_idct_16x16_10_neon
void ff_hevc_idct_16x16_10_neon(int16_t *coeffs, int col_limit)
src
INIT_CLIP pixel * src
Definition: h264pred_template.c:418
ff_hevc_h_loop_filter_chroma_10_neon
void ff_hevc_h_loop_filter_chroma_10_neon(uint8_t *_pix, ptrdiff_t _stride, const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q)
ff_hevc_add_residual_4x4_8_neon
void ff_hevc_add_residual_4x4_8_neon(uint8_t *_dst, const 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, const int16_t *coeffs, ptrdiff_t stride)
NEON8_FNPROTO_PARTIAL_4
#define NEON8_FNPROTO_PARTIAL_4(fn, args, ext)
Definition: hevcdsp_init_aarch64.c:142
ff_hevc_add_residual_8x8_10_neon
void ff_hevc_add_residual_8x8_10_neon(uint8_t *_dst, const int16_t *coeffs, ptrdiff_t stride)
ff_hevc_h_loop_filter_chroma_8_neon
void ff_hevc_h_loop_filter_chroma_8_neon(uint8_t *_pix, ptrdiff_t _stride, const int *_tc, const uint8_t *_no_p, const uint8_t *_no_q)
ff_hevc_add_residual_16x16_8_neon
void ff_hevc_add_residual_16x16_8_neon(uint8_t *_dst, const int16_t *coeffs, ptrdiff_t stride)
NEON8_FNASSIGN_PARTIAL_4
#define NEON8_FNASSIGN_PARTIAL_4(member, v, h, fn, ext)
Definition: hevcdsp_init_aarch64.c:229
ff_hevc_put_hevc_qpel_bi_h16_8_neon
void ff_hevc_put_hevc_qpel_bi_h16_8_neon(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width)
ff_hevc_add_residual_4x4_12_neon
void ff_hevc_add_residual_4x4_12_neon(uint8_t *_dst, const int16_t *coeffs, ptrdiff_t stride)
Generated on Thu Apr 18 2024 22:41:39 for FFmpeg by   doxygen 1.8.17