dsputil: remove shift parameter from scalarproduct_int16
[ffmpeg.git] / libavcodec / arm / dsputil_init_neon.c
1 /*
2  * ARM NEON optimised DSP functions
3  * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
4  *
5  * This file is part of Libav.
6  *
7  * Libav 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  * Libav 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 Libav; 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 "libavcodec/avcodec.h"
25 #include "libavcodec/dsputil.h"
26 #include "dsputil_arm.h"
27
28 void ff_simple_idct_neon(DCTELEM *data);
29 void ff_simple_idct_put_neon(uint8_t *dest, int line_size, DCTELEM *data);
30 void ff_simple_idct_add_neon(uint8_t *dest, int line_size, DCTELEM *data);
31
32 void ff_vp3_idct_neon(DCTELEM *data);
33 void ff_vp3_idct_put_neon(uint8_t *dest, int line_size, DCTELEM *data);
34 void ff_vp3_idct_add_neon(uint8_t *dest, int line_size, DCTELEM *data);
35 void ff_vp3_idct_dc_add_neon(uint8_t *dest, int line_size, const DCTELEM *data);
36
37 void ff_clear_block_neon(DCTELEM *block);
38 void ff_clear_blocks_neon(DCTELEM *blocks);
39
40 void ff_put_pixels16_neon(uint8_t *, const uint8_t *, int, int);
41 void ff_put_pixels16_x2_neon(uint8_t *, const uint8_t *, int, int);
42 void ff_put_pixels16_y2_neon(uint8_t *, const uint8_t *, int, int);
43 void ff_put_pixels16_xy2_neon(uint8_t *, const uint8_t *, int, int);
44 void ff_put_pixels8_neon(uint8_t *, const uint8_t *, int, int);
45 void ff_put_pixels8_x2_neon(uint8_t *, const uint8_t *, int, int);
46 void ff_put_pixels8_y2_neon(uint8_t *, const uint8_t *, int, int);
47 void ff_put_pixels8_xy2_neon(uint8_t *, const uint8_t *, int, int);
48 void ff_put_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
49 void ff_put_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
50 void ff_put_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
51 void ff_put_pixels8_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
52 void ff_put_pixels8_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
53 void ff_put_pixels8_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
54
55 void ff_avg_pixels16_neon(uint8_t *, const uint8_t *, int, int);
56 void ff_avg_pixels16_x2_neon(uint8_t *, const uint8_t *, int, int);
57 void ff_avg_pixels16_y2_neon(uint8_t *, const uint8_t *, int, int);
58 void ff_avg_pixels16_xy2_neon(uint8_t *, const uint8_t *, int, int);
59 void ff_avg_pixels8_neon(uint8_t *, const uint8_t *, int, int);
60 void ff_avg_pixels8_x2_neon(uint8_t *, const uint8_t *, int, int);
61 void ff_avg_pixels8_y2_neon(uint8_t *, const uint8_t *, int, int);
62 void ff_avg_pixels8_xy2_neon(uint8_t *, const uint8_t *, int, int);
63 void ff_avg_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
64 void ff_avg_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
65 void ff_avg_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
66 void ff_avg_pixels8_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
67 void ff_avg_pixels8_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
68 void ff_avg_pixels8_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
69
70 void ff_add_pixels_clamped_neon(const DCTELEM *, uint8_t *, int);
71 void ff_put_pixels_clamped_neon(const DCTELEM *, uint8_t *, int);
72 void ff_put_signed_pixels_clamped_neon(const DCTELEM *, uint8_t *, int);
73
74 void ff_put_h264_qpel16_mc00_neon(uint8_t *, uint8_t *, int);
75 void ff_put_h264_qpel16_mc10_neon(uint8_t *, uint8_t *, int);
76 void ff_put_h264_qpel16_mc20_neon(uint8_t *, uint8_t *, int);
77 void ff_put_h264_qpel16_mc30_neon(uint8_t *, uint8_t *, int);
78 void ff_put_h264_qpel16_mc01_neon(uint8_t *, uint8_t *, int);
79 void ff_put_h264_qpel16_mc11_neon(uint8_t *, uint8_t *, int);
80 void ff_put_h264_qpel16_mc21_neon(uint8_t *, uint8_t *, int);
81 void ff_put_h264_qpel16_mc31_neon(uint8_t *, uint8_t *, int);
82 void ff_put_h264_qpel16_mc02_neon(uint8_t *, uint8_t *, int);
83 void ff_put_h264_qpel16_mc12_neon(uint8_t *, uint8_t *, int);
84 void ff_put_h264_qpel16_mc22_neon(uint8_t *, uint8_t *, int);
85 void ff_put_h264_qpel16_mc32_neon(uint8_t *, uint8_t *, int);
86 void ff_put_h264_qpel16_mc03_neon(uint8_t *, uint8_t *, int);
87 void ff_put_h264_qpel16_mc13_neon(uint8_t *, uint8_t *, int);
88 void ff_put_h264_qpel16_mc23_neon(uint8_t *, uint8_t *, int);
89 void ff_put_h264_qpel16_mc33_neon(uint8_t *, uint8_t *, int);
90
91 void ff_put_h264_qpel8_mc00_neon(uint8_t *, uint8_t *, int);
92 void ff_put_h264_qpel8_mc10_neon(uint8_t *, uint8_t *, int);
93 void ff_put_h264_qpel8_mc20_neon(uint8_t *, uint8_t *, int);
94 void ff_put_h264_qpel8_mc30_neon(uint8_t *, uint8_t *, int);
95 void ff_put_h264_qpel8_mc01_neon(uint8_t *, uint8_t *, int);
96 void ff_put_h264_qpel8_mc11_neon(uint8_t *, uint8_t *, int);
97 void ff_put_h264_qpel8_mc21_neon(uint8_t *, uint8_t *, int);
98 void ff_put_h264_qpel8_mc31_neon(uint8_t *, uint8_t *, int);
99 void ff_put_h264_qpel8_mc02_neon(uint8_t *, uint8_t *, int);
100 void ff_put_h264_qpel8_mc12_neon(uint8_t *, uint8_t *, int);
101 void ff_put_h264_qpel8_mc22_neon(uint8_t *, uint8_t *, int);
102 void ff_put_h264_qpel8_mc32_neon(uint8_t *, uint8_t *, int);
103 void ff_put_h264_qpel8_mc03_neon(uint8_t *, uint8_t *, int);
104 void ff_put_h264_qpel8_mc13_neon(uint8_t *, uint8_t *, int);
105 void ff_put_h264_qpel8_mc23_neon(uint8_t *, uint8_t *, int);
106 void ff_put_h264_qpel8_mc33_neon(uint8_t *, uint8_t *, int);
107
108 void ff_avg_h264_qpel16_mc00_neon(uint8_t *, uint8_t *, int);
109 void ff_avg_h264_qpel16_mc10_neon(uint8_t *, uint8_t *, int);
110 void ff_avg_h264_qpel16_mc20_neon(uint8_t *, uint8_t *, int);
111 void ff_avg_h264_qpel16_mc30_neon(uint8_t *, uint8_t *, int);
112 void ff_avg_h264_qpel16_mc01_neon(uint8_t *, uint8_t *, int);
113 void ff_avg_h264_qpel16_mc11_neon(uint8_t *, uint8_t *, int);
114 void ff_avg_h264_qpel16_mc21_neon(uint8_t *, uint8_t *, int);
115 void ff_avg_h264_qpel16_mc31_neon(uint8_t *, uint8_t *, int);
116 void ff_avg_h264_qpel16_mc02_neon(uint8_t *, uint8_t *, int);
117 void ff_avg_h264_qpel16_mc12_neon(uint8_t *, uint8_t *, int);
118 void ff_avg_h264_qpel16_mc22_neon(uint8_t *, uint8_t *, int);
119 void ff_avg_h264_qpel16_mc32_neon(uint8_t *, uint8_t *, int);
120 void ff_avg_h264_qpel16_mc03_neon(uint8_t *, uint8_t *, int);
121 void ff_avg_h264_qpel16_mc13_neon(uint8_t *, uint8_t *, int);
122 void ff_avg_h264_qpel16_mc23_neon(uint8_t *, uint8_t *, int);
123 void ff_avg_h264_qpel16_mc33_neon(uint8_t *, uint8_t *, int);
124
125 void ff_avg_h264_qpel8_mc00_neon(uint8_t *, uint8_t *, int);
126 void ff_avg_h264_qpel8_mc10_neon(uint8_t *, uint8_t *, int);
127 void ff_avg_h264_qpel8_mc20_neon(uint8_t *, uint8_t *, int);
128 void ff_avg_h264_qpel8_mc30_neon(uint8_t *, uint8_t *, int);
129 void ff_avg_h264_qpel8_mc01_neon(uint8_t *, uint8_t *, int);
130 void ff_avg_h264_qpel8_mc11_neon(uint8_t *, uint8_t *, int);
131 void ff_avg_h264_qpel8_mc21_neon(uint8_t *, uint8_t *, int);
132 void ff_avg_h264_qpel8_mc31_neon(uint8_t *, uint8_t *, int);
133 void ff_avg_h264_qpel8_mc02_neon(uint8_t *, uint8_t *, int);
134 void ff_avg_h264_qpel8_mc12_neon(uint8_t *, uint8_t *, int);
135 void ff_avg_h264_qpel8_mc22_neon(uint8_t *, uint8_t *, int);
136 void ff_avg_h264_qpel8_mc32_neon(uint8_t *, uint8_t *, int);
137 void ff_avg_h264_qpel8_mc03_neon(uint8_t *, uint8_t *, int);
138 void ff_avg_h264_qpel8_mc13_neon(uint8_t *, uint8_t *, int);
139 void ff_avg_h264_qpel8_mc23_neon(uint8_t *, uint8_t *, int);
140 void ff_avg_h264_qpel8_mc33_neon(uint8_t *, uint8_t *, int);
141
142 void ff_put_h264_chroma_mc8_neon(uint8_t *, uint8_t *, int, int, int, int);
143 void ff_put_h264_chroma_mc4_neon(uint8_t *, uint8_t *, int, int, int, int);
144 void ff_put_h264_chroma_mc2_neon(uint8_t *, uint8_t *, int, int, int, int);
145
146 void ff_avg_h264_chroma_mc8_neon(uint8_t *, uint8_t *, int, int, int, int);
147 void ff_avg_h264_chroma_mc4_neon(uint8_t *, uint8_t *, int, int, int, int);
148 void ff_avg_h264_chroma_mc2_neon(uint8_t *, uint8_t *, int, int, int, int);
149
150 void ff_vp3_v_loop_filter_neon(uint8_t *, int, int *);
151 void ff_vp3_h_loop_filter_neon(uint8_t *, int, int *);
152
153 void ff_vector_fmul_neon(float *dst, const float *src0, const float *src1, int len);
154 void ff_vector_fmul_window_neon(float *dst, const float *src0,
155                                 const float *src1, const float *win, int len);
156 void ff_vector_fmul_scalar_neon(float *dst, const float *src, float mul,
157                                 int len);
158 void ff_vector_fmac_scalar_neon(float *dst, const float *src, float mul,
159                                 int len);
160 void ff_butterflies_float_neon(float *v1, float *v2, int len);
161 float ff_scalarproduct_float_neon(const float *v1, const float *v2, int len);
162 void ff_vector_fmul_reverse_neon(float *dst, const float *src0,
163                                  const float *src1, int len);
164 void ff_vector_fmul_add_neon(float *dst, const float *src0, const float *src1,
165                              const float *src2, int len);
166
167 void ff_vector_clipf_neon(float *dst, const float *src, float min, float max,
168                           int len);
169 void ff_vector_clip_int32_neon(int32_t *dst, const int32_t *src, int32_t min,
170                                int32_t max, unsigned int len);
171
172 void ff_vorbis_inverse_coupling_neon(float *mag, float *ang, int blocksize);
173
174 int32_t ff_scalarproduct_int16_neon(const int16_t *v1, const int16_t *v2, int len);
175 int32_t ff_scalarproduct_and_madd_int16_neon(int16_t *v1, const int16_t *v2,
176                                              const int16_t *v3, int len, int mul);
177
178 void ff_apply_window_int16_neon(int16_t *dst, const int16_t *src,
179                                 const int16_t *window, unsigned n);
180
181 void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
182 {
183     const int high_bit_depth = avctx->bits_per_raw_sample > 8;
184
185     if (!avctx->lowres && avctx->bits_per_raw_sample <= 8) {
186         if (avctx->idct_algo == FF_IDCT_AUTO ||
187             avctx->idct_algo == FF_IDCT_SIMPLENEON) {
188             c->idct_put              = ff_simple_idct_put_neon;
189             c->idct_add              = ff_simple_idct_add_neon;
190             c->idct                  = ff_simple_idct_neon;
191             c->idct_permutation_type = FF_PARTTRANS_IDCT_PERM;
192         } else if ((CONFIG_VP3_DECODER || CONFIG_VP5_DECODER ||
193                     CONFIG_VP6_DECODER) &&
194                    avctx->idct_algo == FF_IDCT_VP3) {
195             c->idct_put              = ff_vp3_idct_put_neon;
196             c->idct_add              = ff_vp3_idct_add_neon;
197             c->idct                  = ff_vp3_idct_neon;
198             c->idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
199         }
200     }
201
202     if (!high_bit_depth) {
203         c->clear_block  = ff_clear_block_neon;
204         c->clear_blocks = ff_clear_blocks_neon;
205
206         c->put_pixels_tab[0][0] = ff_put_pixels16_neon;
207         c->put_pixels_tab[0][1] = ff_put_pixels16_x2_neon;
208         c->put_pixels_tab[0][2] = ff_put_pixels16_y2_neon;
209         c->put_pixels_tab[0][3] = ff_put_pixels16_xy2_neon;
210         c->put_pixels_tab[1][0] = ff_put_pixels8_neon;
211         c->put_pixels_tab[1][1] = ff_put_pixels8_x2_neon;
212         c->put_pixels_tab[1][2] = ff_put_pixels8_y2_neon;
213         c->put_pixels_tab[1][3] = ff_put_pixels8_xy2_neon;
214
215         c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_neon;
216         c->put_no_rnd_pixels_tab[0][1] = ff_put_pixels16_x2_no_rnd_neon;
217         c->put_no_rnd_pixels_tab[0][2] = ff_put_pixels16_y2_no_rnd_neon;
218         c->put_no_rnd_pixels_tab[0][3] = ff_put_pixels16_xy2_no_rnd_neon;
219         c->put_no_rnd_pixels_tab[1][0] = ff_put_pixels8_neon;
220         c->put_no_rnd_pixels_tab[1][1] = ff_put_pixels8_x2_no_rnd_neon;
221         c->put_no_rnd_pixels_tab[1][2] = ff_put_pixels8_y2_no_rnd_neon;
222         c->put_no_rnd_pixels_tab[1][3] = ff_put_pixels8_xy2_no_rnd_neon;
223
224         c->avg_pixels_tab[0][0] = ff_avg_pixels16_neon;
225         c->avg_pixels_tab[0][1] = ff_avg_pixels16_x2_neon;
226         c->avg_pixels_tab[0][2] = ff_avg_pixels16_y2_neon;
227         c->avg_pixels_tab[0][3] = ff_avg_pixels16_xy2_neon;
228         c->avg_pixels_tab[1][0] = ff_avg_pixels8_neon;
229         c->avg_pixels_tab[1][1] = ff_avg_pixels8_x2_neon;
230         c->avg_pixels_tab[1][2] = ff_avg_pixels8_y2_neon;
231         c->avg_pixels_tab[1][3] = ff_avg_pixels8_xy2_neon;
232
233         c->avg_no_rnd_pixels_tab[0][0] = ff_avg_pixels16_neon;
234         c->avg_no_rnd_pixels_tab[0][1] = ff_avg_pixels16_x2_no_rnd_neon;
235         c->avg_no_rnd_pixels_tab[0][2] = ff_avg_pixels16_y2_no_rnd_neon;
236         c->avg_no_rnd_pixels_tab[0][3] = ff_avg_pixels16_xy2_no_rnd_neon;
237         c->avg_no_rnd_pixels_tab[1][0] = ff_avg_pixels8_neon;
238         c->avg_no_rnd_pixels_tab[1][1] = ff_avg_pixels8_x2_no_rnd_neon;
239         c->avg_no_rnd_pixels_tab[1][2] = ff_avg_pixels8_y2_no_rnd_neon;
240         c->avg_no_rnd_pixels_tab[1][3] = ff_avg_pixels8_xy2_no_rnd_neon;
241     }
242
243     c->add_pixels_clamped = ff_add_pixels_clamped_neon;
244     c->put_pixels_clamped = ff_put_pixels_clamped_neon;
245     c->put_signed_pixels_clamped = ff_put_signed_pixels_clamped_neon;
246
247     if (CONFIG_H264_DECODER && !high_bit_depth) {
248         c->put_h264_chroma_pixels_tab[0] = ff_put_h264_chroma_mc8_neon;
249         c->put_h264_chroma_pixels_tab[1] = ff_put_h264_chroma_mc4_neon;
250         c->put_h264_chroma_pixels_tab[2] = ff_put_h264_chroma_mc2_neon;
251
252         c->avg_h264_chroma_pixels_tab[0] = ff_avg_h264_chroma_mc8_neon;
253         c->avg_h264_chroma_pixels_tab[1] = ff_avg_h264_chroma_mc4_neon;
254         c->avg_h264_chroma_pixels_tab[2] = ff_avg_h264_chroma_mc2_neon;
255
256         c->put_h264_qpel_pixels_tab[0][ 0] = ff_put_h264_qpel16_mc00_neon;
257         c->put_h264_qpel_pixels_tab[0][ 1] = ff_put_h264_qpel16_mc10_neon;
258         c->put_h264_qpel_pixels_tab[0][ 2] = ff_put_h264_qpel16_mc20_neon;
259         c->put_h264_qpel_pixels_tab[0][ 3] = ff_put_h264_qpel16_mc30_neon;
260         c->put_h264_qpel_pixels_tab[0][ 4] = ff_put_h264_qpel16_mc01_neon;
261         c->put_h264_qpel_pixels_tab[0][ 5] = ff_put_h264_qpel16_mc11_neon;
262         c->put_h264_qpel_pixels_tab[0][ 6] = ff_put_h264_qpel16_mc21_neon;
263         c->put_h264_qpel_pixels_tab[0][ 7] = ff_put_h264_qpel16_mc31_neon;
264         c->put_h264_qpel_pixels_tab[0][ 8] = ff_put_h264_qpel16_mc02_neon;
265         c->put_h264_qpel_pixels_tab[0][ 9] = ff_put_h264_qpel16_mc12_neon;
266         c->put_h264_qpel_pixels_tab[0][10] = ff_put_h264_qpel16_mc22_neon;
267         c->put_h264_qpel_pixels_tab[0][11] = ff_put_h264_qpel16_mc32_neon;
268         c->put_h264_qpel_pixels_tab[0][12] = ff_put_h264_qpel16_mc03_neon;
269         c->put_h264_qpel_pixels_tab[0][13] = ff_put_h264_qpel16_mc13_neon;
270         c->put_h264_qpel_pixels_tab[0][14] = ff_put_h264_qpel16_mc23_neon;
271         c->put_h264_qpel_pixels_tab[0][15] = ff_put_h264_qpel16_mc33_neon;
272
273         c->put_h264_qpel_pixels_tab[1][ 0] = ff_put_h264_qpel8_mc00_neon;
274         c->put_h264_qpel_pixels_tab[1][ 1] = ff_put_h264_qpel8_mc10_neon;
275         c->put_h264_qpel_pixels_tab[1][ 2] = ff_put_h264_qpel8_mc20_neon;
276         c->put_h264_qpel_pixels_tab[1][ 3] = ff_put_h264_qpel8_mc30_neon;
277         c->put_h264_qpel_pixels_tab[1][ 4] = ff_put_h264_qpel8_mc01_neon;
278         c->put_h264_qpel_pixels_tab[1][ 5] = ff_put_h264_qpel8_mc11_neon;
279         c->put_h264_qpel_pixels_tab[1][ 6] = ff_put_h264_qpel8_mc21_neon;
280         c->put_h264_qpel_pixels_tab[1][ 7] = ff_put_h264_qpel8_mc31_neon;
281         c->put_h264_qpel_pixels_tab[1][ 8] = ff_put_h264_qpel8_mc02_neon;
282         c->put_h264_qpel_pixels_tab[1][ 9] = ff_put_h264_qpel8_mc12_neon;
283         c->put_h264_qpel_pixels_tab[1][10] = ff_put_h264_qpel8_mc22_neon;
284         c->put_h264_qpel_pixels_tab[1][11] = ff_put_h264_qpel8_mc32_neon;
285         c->put_h264_qpel_pixels_tab[1][12] = ff_put_h264_qpel8_mc03_neon;
286         c->put_h264_qpel_pixels_tab[1][13] = ff_put_h264_qpel8_mc13_neon;
287         c->put_h264_qpel_pixels_tab[1][14] = ff_put_h264_qpel8_mc23_neon;
288         c->put_h264_qpel_pixels_tab[1][15] = ff_put_h264_qpel8_mc33_neon;
289
290         c->avg_h264_qpel_pixels_tab[0][ 0] = ff_avg_h264_qpel16_mc00_neon;
291         c->avg_h264_qpel_pixels_tab[0][ 1] = ff_avg_h264_qpel16_mc10_neon;
292         c->avg_h264_qpel_pixels_tab[0][ 2] = ff_avg_h264_qpel16_mc20_neon;
293         c->avg_h264_qpel_pixels_tab[0][ 3] = ff_avg_h264_qpel16_mc30_neon;
294         c->avg_h264_qpel_pixels_tab[0][ 4] = ff_avg_h264_qpel16_mc01_neon;
295         c->avg_h264_qpel_pixels_tab[0][ 5] = ff_avg_h264_qpel16_mc11_neon;
296         c->avg_h264_qpel_pixels_tab[0][ 6] = ff_avg_h264_qpel16_mc21_neon;
297         c->avg_h264_qpel_pixels_tab[0][ 7] = ff_avg_h264_qpel16_mc31_neon;
298         c->avg_h264_qpel_pixels_tab[0][ 8] = ff_avg_h264_qpel16_mc02_neon;
299         c->avg_h264_qpel_pixels_tab[0][ 9] = ff_avg_h264_qpel16_mc12_neon;
300         c->avg_h264_qpel_pixels_tab[0][10] = ff_avg_h264_qpel16_mc22_neon;
301         c->avg_h264_qpel_pixels_tab[0][11] = ff_avg_h264_qpel16_mc32_neon;
302         c->avg_h264_qpel_pixels_tab[0][12] = ff_avg_h264_qpel16_mc03_neon;
303         c->avg_h264_qpel_pixels_tab[0][13] = ff_avg_h264_qpel16_mc13_neon;
304         c->avg_h264_qpel_pixels_tab[0][14] = ff_avg_h264_qpel16_mc23_neon;
305         c->avg_h264_qpel_pixels_tab[0][15] = ff_avg_h264_qpel16_mc33_neon;
306
307         c->avg_h264_qpel_pixels_tab[1][ 0] = ff_avg_h264_qpel8_mc00_neon;
308         c->avg_h264_qpel_pixels_tab[1][ 1] = ff_avg_h264_qpel8_mc10_neon;
309         c->avg_h264_qpel_pixels_tab[1][ 2] = ff_avg_h264_qpel8_mc20_neon;
310         c->avg_h264_qpel_pixels_tab[1][ 3] = ff_avg_h264_qpel8_mc30_neon;
311         c->avg_h264_qpel_pixels_tab[1][ 4] = ff_avg_h264_qpel8_mc01_neon;
312         c->avg_h264_qpel_pixels_tab[1][ 5] = ff_avg_h264_qpel8_mc11_neon;
313         c->avg_h264_qpel_pixels_tab[1][ 6] = ff_avg_h264_qpel8_mc21_neon;
314         c->avg_h264_qpel_pixels_tab[1][ 7] = ff_avg_h264_qpel8_mc31_neon;
315         c->avg_h264_qpel_pixels_tab[1][ 8] = ff_avg_h264_qpel8_mc02_neon;
316         c->avg_h264_qpel_pixels_tab[1][ 9] = ff_avg_h264_qpel8_mc12_neon;
317         c->avg_h264_qpel_pixels_tab[1][10] = ff_avg_h264_qpel8_mc22_neon;
318         c->avg_h264_qpel_pixels_tab[1][11] = ff_avg_h264_qpel8_mc32_neon;
319         c->avg_h264_qpel_pixels_tab[1][12] = ff_avg_h264_qpel8_mc03_neon;
320         c->avg_h264_qpel_pixels_tab[1][13] = ff_avg_h264_qpel8_mc13_neon;
321         c->avg_h264_qpel_pixels_tab[1][14] = ff_avg_h264_qpel8_mc23_neon;
322         c->avg_h264_qpel_pixels_tab[1][15] = ff_avg_h264_qpel8_mc33_neon;
323     }
324
325     if (CONFIG_VP3_DECODER) {
326         c->vp3_v_loop_filter = ff_vp3_v_loop_filter_neon;
327         c->vp3_h_loop_filter = ff_vp3_h_loop_filter_neon;
328         c->vp3_idct_dc_add   = ff_vp3_idct_dc_add_neon;
329     }
330
331     c->vector_fmul                = ff_vector_fmul_neon;
332     c->vector_fmul_window         = ff_vector_fmul_window_neon;
333     c->vector_fmul_scalar         = ff_vector_fmul_scalar_neon;
334     c->vector_fmac_scalar         = ff_vector_fmac_scalar_neon;
335     c->butterflies_float          = ff_butterflies_float_neon;
336     c->scalarproduct_float        = ff_scalarproduct_float_neon;
337     c->vector_fmul_reverse        = ff_vector_fmul_reverse_neon;
338     c->vector_fmul_add            = ff_vector_fmul_add_neon;
339     c->vector_clipf               = ff_vector_clipf_neon;
340     c->vector_clip_int32          = ff_vector_clip_int32_neon;
341
342     if (CONFIG_VORBIS_DECODER)
343         c->vorbis_inverse_coupling = ff_vorbis_inverse_coupling_neon;
344
345     c->scalarproduct_int16 = ff_scalarproduct_int16_neon;
346     c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_neon;
347
348     c->apply_window_int16 = ff_apply_window_int16_neon;
349 }