GCC Code Coverage Report
Directory: ../../../ffmpeg/ Exec Total Coverage
File: src/libavcodec/hevcpred_template.c Lines: 305 312 97.8 %
Date: 2021-04-18 10:33:33 Branches: 290 311 93.2 %

Line Branch Exec Source
1
/*
2
 * HEVC video decoder
3
 *
4
 * Copyright (C) 2012 - 2013 Guillaume Martres
5
 *
6
 * This file is part of FFmpeg.
7
 *
8
 * FFmpeg is free software; you can redistribute it and/or
9
 * modify it under the terms of the GNU Lesser General Public
10
 * License as published by the Free Software Foundation; either
11
 * version 2.1 of the License, or (at your option) any later version.
12
 *
13
 * FFmpeg is distributed in the hope that it will be useful,
14
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
16
 * Lesser General Public License for more details.
17
 *
18
 * You should have received a copy of the GNU Lesser General Public
19
 * License along with FFmpeg; if not, write to the Free Software
20
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21
 */
22
23
#include "libavutil/pixdesc.h"
24
25
#include "bit_depth_template.c"
26
#include "hevcpred.h"
27
28
#define POS(x, y) src[(x) + stride * (y)]
29
30
41176846
static av_always_inline void FUNC(intra_pred)(HEVCContext *s, int x0, int y0,
31
                                              int log2_size, int c_idx)
32
{
33
#define PU(x) \
34
    ((x) >> s->ps.sps->log2_min_pu_size)
35
#define MVF(x, y) \
36
    (s->ref->tab_mvf[(x) + (y) * min_pu_width])
37
#define MVF_PU(x, y) \
38
    MVF(PU(x0 + ((x) * (1 << hshift))), PU(y0 + ((y) * (1 << vshift))))
39
#define IS_INTRA(x, y) \
40
    (MVF_PU(x, y).pred_flag == PF_INTRA)
41
#define MIN_TB_ADDR_ZS(x, y) \
42
    s->ps.pps->min_tb_addr_zs[(y) * (s->ps.sps->tb_mask+2) + (x)]
43
#define EXTEND(ptr, val, len)         \
44
do {                                  \
45
    pixel4 pix = PIXEL_SPLAT_X4(val); \
46
    for (i = 0; i < (len); i += 4)    \
47
        AV_WN4P(ptr + i, pix);        \
48
} while (0)
49
50
#define EXTEND_RIGHT_CIP(ptr, start, length)                                   \
51
        for (i = start; i < (start) + (length); i += 4)                        \
52
            if (!IS_INTRA(i, -1))                                              \
53
                AV_WN4P(&ptr[i], a);                                           \
54
            else                                                               \
55
                a = PIXEL_SPLAT_X4(ptr[i+3])
56
#define EXTEND_LEFT_CIP(ptr, start, length) \
57
        for (i = start; i > (start) - (length); i--) \
58
            if (!IS_INTRA(i - 1, -1)) \
59
                ptr[i - 1] = ptr[i]
60
#define EXTEND_UP_CIP(ptr, start, length)                                      \
61
        for (i = (start); i > (start) - (length); i -= 4)                      \
62
            if (!IS_INTRA(-1, i - 3))                                          \
63
                AV_WN4P(&ptr[i - 3], a);                                       \
64
            else                                                               \
65
                a = PIXEL_SPLAT_X4(ptr[i - 3])
66
#define EXTEND_DOWN_CIP(ptr, start, length)                                    \
67
        for (i = start; i < (start) + (length); i += 4)                        \
68
            if (!IS_INTRA(-1, i))                                              \
69
                AV_WN4P(&ptr[i], a);                                           \
70
            else                                                               \
71
                a = PIXEL_SPLAT_X4(ptr[i + 3])
72
73
41176846
    HEVCLocalContext *lc = s->HEVClc;
74
    int i;
75
41176846
    int hshift = s->ps.sps->hshift[c_idx];
76
41176846
    int vshift = s->ps.sps->vshift[c_idx];
77
41176846
    int size = (1 << log2_size);
78
41176846
    int size_in_luma_h = size << hshift;
79
41176846
    int size_in_tbs_h  = size_in_luma_h >> s->ps.sps->log2_min_tb_size;
80
41176846
    int size_in_luma_v = size << vshift;
81
41176846
    int size_in_tbs_v  = size_in_luma_v >> s->ps.sps->log2_min_tb_size;
82
41176846
    int x = x0 >> hshift;
83
41176846
    int y = y0 >> vshift;
84
41176846
    int x_tb = (x0 >> s->ps.sps->log2_min_tb_size) & s->ps.sps->tb_mask;
85
41176846
    int y_tb = (y0 >> s->ps.sps->log2_min_tb_size) & s->ps.sps->tb_mask;
86

41176846
    int spin = c_idx && !size_in_tbs_v && ((2 * y0) & (1 << s->ps.sps->log2_min_tb_size));
87
88
41176846
    int cur_tb_addr = MIN_TB_ADDR_ZS(x_tb, y_tb);
89
90
41176846
    ptrdiff_t stride = s->frame->linesize[c_idx] / sizeof(pixel);
91
41176846
    pixel *src = (pixel*)s->frame->data[c_idx] + x + y * stride;
92
93
41176846
    int min_pu_width = s->ps.sps->min_pu_width;
94
95
41176846
    enum IntraPredMode mode = c_idx ? lc->tu.intra_pred_mode_c :
96
19161598
                              lc->tu.intra_pred_mode;
97
    pixel4 a;
98
    pixel  left_array[2 * MAX_TB_SIZE + 1];
99
    pixel  filtered_left_array[2 * MAX_TB_SIZE + 1];
100
    pixel  top_array[2 * MAX_TB_SIZE + 1];
101
    pixel  filtered_top_array[2 * MAX_TB_SIZE + 1];
102
103
41176846
    pixel  *left          = left_array + 1;
104
41176846
    pixel  *top           = top_array  + 1;
105
41176846
    pixel  *filtered_left = filtered_left_array + 1;
106
41176846
    pixel  *filtered_top  = filtered_top_array  + 1;
107

41176846
    int cand_bottom_left = lc->na.cand_bottom_left && cur_tb_addr > MIN_TB_ADDR_ZS( x_tb - 1, (y_tb + size_in_tbs_v + spin) & s->ps.sps->tb_mask);
108
41176846
    int cand_left        = lc->na.cand_left;
109
41176846
    int cand_up_left     = lc->na.cand_up_left;
110
41176846
    int cand_up          = lc->na.cand_up;
111

41176846
    int cand_up_right    = lc->na.cand_up_right && !spin && cur_tb_addr > MIN_TB_ADDR_ZS((x_tb + size_in_tbs_h) & s->ps.sps->tb_mask, y_tb - 1);
112
113
41176846
    int bottom_left_size = (FFMIN(y0 + 2 * size_in_luma_v, s->ps.sps->height) -
114
41176846
                           (y0 + size_in_luma_v)) >> vshift;
115
41176846
    int top_right_size   = (FFMIN(x0 + 2 * size_in_luma_h, s->ps.sps->width) -
116
41176846
                           (x0 + size_in_luma_h)) >> hshift;
117
118
41176846
    if (s->ps.pps->constrained_intra_pred_flag == 1) {
119
271050
        int size_in_luma_pu_v = PU(size_in_luma_v);
120
271050
        int size_in_luma_pu_h = PU(size_in_luma_h);
121
271050
        int on_pu_edge_x    = !av_mod_uintp2(x0, s->ps.sps->log2_min_pu_size);
122
271050
        int on_pu_edge_y    = !av_mod_uintp2(y0, s->ps.sps->log2_min_pu_size);
123
271050
        if (!size_in_luma_pu_h)
124
            size_in_luma_pu_h++;
125

271050
        if (cand_bottom_left == 1 && on_pu_edge_x) {
126
84160
            int x_left_pu   = PU(x0 - 1);
127
84160
            int y_bottom_pu = PU(y0 + size_in_luma_v);
128
84160
            int max = FFMIN(size_in_luma_pu_v, s->ps.sps->min_pu_height - y_bottom_pu);
129
84160
            cand_bottom_left = 0;
130
179804
            for (i = 0; i < max; i += 2)
131
95644
                cand_bottom_left |= (MVF(x_left_pu, y_bottom_pu + i).pred_flag == PF_INTRA);
132
        }
133

271050
        if (cand_left == 1 && on_pu_edge_x) {
134
266214
            int x_left_pu   = PU(x0 - 1);
135
266214
            int y_left_pu   = PU(y0);
136
266214
            int max = FFMIN(size_in_luma_pu_v, s->ps.sps->min_pu_height - y_left_pu);
137
266214
            cand_left = 0;
138
574872
            for (i = 0; i < max; i += 2)
139
308658
                cand_left |= (MVF(x_left_pu, y_left_pu + i).pred_flag == PF_INTRA);
140
        }
141
271050
        if (cand_up_left == 1) {
142
258168
            int x_left_pu   = PU(x0 - 1);
143
258168
            int y_top_pu    = PU(y0 - 1);
144
258168
            cand_up_left = MVF(x_left_pu, y_top_pu).pred_flag == PF_INTRA;
145
        }
146

271050
        if (cand_up == 1 && on_pu_edge_y) {
147
262806
            int x_top_pu    = PU(x0);
148
262806
            int y_top_pu    = PU(y0 - 1);
149
262806
            int max = FFMIN(size_in_luma_pu_h, s->ps.sps->min_pu_width - x_top_pu);
150
262806
            cand_up = 0;
151
565704
            for (i = 0; i < max; i += 2)
152
302898
                cand_up |= (MVF(x_top_pu + i, y_top_pu).pred_flag == PF_INTRA);
153
        }
154

271050
        if (cand_up_right == 1 && on_pu_edge_y) {
155
174386
            int y_top_pu    = PU(y0 - 1);
156
174386
            int x_right_pu  = PU(x0 + size_in_luma_h);
157
174386
            int max = FFMIN(size_in_luma_pu_h, s->ps.sps->min_pu_width - x_right_pu);
158
174386
            cand_up_right = 0;
159
375976
            for (i = 0; i < max; i += 2)
160
201590
                cand_up_right |= (MVF(x_right_pu + i, y_top_pu).pred_flag == PF_INTRA);
161
        }
162
271050
        memset(left, 128, 2 * MAX_TB_SIZE*sizeof(pixel));
163
271050
        memset(top , 128, 2 * MAX_TB_SIZE*sizeof(pixel));
164
271050
        top[-1] = 128;
165
    }
166
41176846
    if (cand_up_left) {
167
39963310
        left[-1] = POS(-1, -1);
168
39963310
        top[-1]  = left[-1];
169
    }
170
41176846
    if (cand_up)
171
40426896
        memcpy(top, src - stride, size * sizeof(pixel));
172
41176846
    if (cand_up_right) {
173
26232882
        memcpy(top + size, src - stride + size, size * sizeof(pixel));
174
26286190
        EXTEND(top + size + top_right_size, POS(size + top_right_size - 1, -1),
175
               size - top_right_size);
176
    }
177
41176846
    if (cand_left)
178
285090982
        for (i = 0; i < size; i++)
179
244401744
            left[i] = POS(-1, i);
180
41176846
    if (cand_bottom_left) {
181
92257320
        for (i = size; i < size + bottom_left_size; i++)
182
78578400
            left[i] = POS(-1, i);
183
13743572
        EXTEND(left + size + bottom_left_size, POS(-1, size + bottom_left_size - 1),
184
               size - bottom_left_size);
185
    }
186
187
41176846
    if (s->ps.pps->constrained_intra_pred_flag == 1) {
188


271050
        if (cand_bottom_left || cand_left || cand_up_left || cand_up || cand_up_right) {
189
535404
            int size_max_x = x0 + ((2 * size) << hshift) < s->ps.sps->width ?
190
267702
                                    2 * size : (s->ps.sps->width - x0) >> hshift;
191
535404
            int size_max_y = y0 + ((2 * size) << vshift) < s->ps.sps->height ?
192
267702
                                    2 * size : (s->ps.sps->height - y0) >> vshift;
193
267702
            int j = size + (cand_bottom_left? bottom_left_size: 0) -1;
194
267702
            if (!cand_up_right) {
195
104394
                size_max_x = x0 + ((size) << hshift) < s->ps.sps->width ?
196
104394
                                                    size : (s->ps.sps->width - x0) >> hshift;
197
            }
198
267702
            if (!cand_bottom_left) {
199
189324
                size_max_y = y0 + (( size) << vshift) < s->ps.sps->height ?
200
189324
                                                     size : (s->ps.sps->height - y0) >> vshift;
201
            }
202

267702
            if (cand_bottom_left || cand_left || cand_up_left) {
203

273404
                while (j > -1 && !IS_INTRA(-1, j))
204
17280
                    j--;
205
256124
                if (!IS_INTRA(-1, j)) {
206
                    j = 0;
207
                    while (j < size_max_x && !IS_INTRA(j, -1))
208
                        j++;
209
                    EXTEND_LEFT_CIP(top, j, j + 1);
210
                    left[-1] = top[-1];
211
                }
212
            } else {
213
11578
                j = 0;
214

18858
                while (j < size_max_x && !IS_INTRA(j, -1))
215
7280
                    j++;
216
11578
                if (j > 0)
217
818
                    if (cand_up_left) {
218
                        EXTEND_LEFT_CIP(top, j, j + 1);
219
                    } else {
220

8098
                        EXTEND_LEFT_CIP(top, j, j);
221
818
                        top[-1] = top[0];
222
                    }
223
11578
                left[-1] = top[-1];
224
            }
225
267702
            left[-1] = top[-1];
226

267702
            if (cand_bottom_left || cand_left) {
227
253796
                a = PIXEL_SPLAT_X4(left[-1]);
228

702646
                EXTEND_DOWN_CIP(left, 0, size_max_y);
229
            }
230
267702
            if (!cand_left)
231
38910
                EXTEND(left, left[-1], size);
232
267702
            if (!cand_bottom_left)
233
453010
                EXTEND(left + size, left[size - 1], size);
234

267702
            if (x0 != 0 && y0 != 0) {
235
260370
                a = PIXEL_SPLAT_X4(left[size_max_y - 1]);
236

714714
                EXTEND_UP_CIP(left, size_max_y - 1, size_max_y);
237
260370
                if (!IS_INTRA(-1, - 1))
238
14852
                    left[-1] = left[0];
239
7332
            } else if (x0 == 0) {
240
8996
                EXTEND(left, 0, size_max_y);
241
            } else {
242
4228
                a = PIXEL_SPLAT_X4(left[size_max_y - 1]);
243

16018
                EXTEND_UP_CIP(left, size_max_y - 1, size_max_y);
244
            }
245
267702
            top[-1] = left[-1];
246
267702
            if (y0 != 0) {
247
263474
                a = PIXEL_SPLAT_X4(left[-1]);
248

844462
                EXTEND_RIGHT_CIP(top, 0, size_max_x);
249
            }
250
        }
251
    }
252
    // Infer the unavailable samples
253
41176846
    if (!cand_bottom_left) {
254
27497926
        if (cand_left) {
255
68403330
            EXTEND(left + size, left[size - 1], size);
256
486912
        } else if (cand_up_left) {
257
9640
            EXTEND(left, left[-1], 2 * size);
258
2328
            cand_left = 1;
259
484584
        } else if (cand_up) {
260
463590
            left[-1] = top[0];
261
2597242
            EXTEND(left, left[-1], 2 * size);
262
463590
            cand_up_left = 1;
263
463590
            cand_left    = 1;
264
20994
        } else if (cand_up_right) {
265
2132
            EXTEND(top, top[size], size);
266
680
            left[-1] = top[size];
267
3584
            EXTEND(left, left[-1], 2 * size);
268
680
            cand_up      = 1;
269
680
            cand_up_left = 1;
270
680
            cand_left    = 1;
271
        } else { // No samples available
272
20314
            left[-1] = (1 << (BIT_DEPTH - 1));
273
134998
            EXTEND(top,  left[-1], 2 * size);
274
134998
            EXTEND(left, left[-1], 2 * size);
275
        }
276
    }
277
278
41176846
    if (!cand_left)
279
79484
        EXTEND(left, left[size], size);
280
41176846
    if (!cand_up_left) {
281
749266
        left[-1] = left[0];
282
    }
283
41176846
    if (!cand_up)
284
2309802
        EXTEND(top, left[-1], size);
285
41176846
    if (!cand_up_right)
286
37193716
        EXTEND(top + size, top[size - 1], size);
287
288
41176846
    top[-1] = left[-1];
289
290
    // Filtering process
291

41176846
    if (!s->ps.sps->intra_smoothing_disabled_flag && (c_idx == 0  || s->ps.sps->chroma_format_idc == 3)) {
292

19613488
        if (mode != INTRA_DC && size != 4){
293
6266918
            int intra_hor_ver_dist_thresh[] = { 7, 1, 0 };
294

6266918
            int min_dist_vert_hor = FFMIN(FFABS((int)(mode - 26U)),
295
                                          FFABS((int)(mode - 10U)));
296
6266918
            if (min_dist_vert_hor > intra_hor_ver_dist_thresh[log2_size - 3]) {
297
2563572
                int threshold = 1 << (BIT_DEPTH - 5);
298

2563572
                if (s->ps.sps->sps_strong_intra_smoothing_enable_flag && c_idx == 0 &&
299
285490
                    log2_size == 5 &&
300
285490
                    FFABS(top[-1]  + top[63]  - 2 * top[31])  < threshold &&
301
111016
                    FFABS(left[-1] + left[63] - 2 * left[31]) < threshold) {
302
                    // We can't just overwrite values in top because it could be
303
                    // a pointer into src
304
60920
                    filtered_top[-1] = top[-1];
305
60920
                    filtered_top[63] = top[63];
306
3898880
                    for (i = 0; i < 63; i++)
307
3837960
                        filtered_top[i] = ((64 - (i + 1)) * top[-1] +
308
3837960
                                           (i + 1)  * top[63] + 32) >> 6;
309
3898880
                    for (i = 0; i < 63; i++)
310
3837960
                        left[i] = ((64 - (i + 1)) * left[-1] +
311
3837960
                                   (i + 1)  * left[63] + 32) >> 6;
312
60920
                    top = filtered_top;
313
                } else {
314
2502652
                    filtered_left[2 * size - 1] = left[2 * size - 1];
315
2502652
                    filtered_top[2 * size - 1]  = top[2 * size - 1];
316
69825184
                    for (i = 2 * size - 2; i >= 0; i--)
317
67322532
                        filtered_left[i] = (left[i + 1] + 2 * left[i] +
318
67322532
                                            left[i - 1] + 2) >> 2;
319
2502652
                    filtered_top[-1]  =
320
2502652
                    filtered_left[-1] = (left[0] + 2 * left[-1] + top[0] + 2) >> 2;
321
69825184
                    for (i = 2 * size - 2; i >= 0; i--)
322
67322532
                        filtered_top[i] = (top[i + 1] + 2 * top[i] +
323
67322532
                                           top[i - 1] + 2) >> 2;
324
2502652
                    left = filtered_left;
325
2502652
                    top  = filtered_top;
326
                }
327
            }
328
        }
329
    }
330
331
41176846
    switch (mode) {
332
8744180
    case INTRA_PLANAR:
333
8744180
        s->hpc.pred_planar[log2_size - 2]((uint8_t *)src, (uint8_t *)top,
334
                                          (uint8_t *)left, stride);
335
8744180
        break;
336
5402678
    case INTRA_DC:
337
5402678
        s->hpc.pred_dc((uint8_t *)src, (uint8_t *)top,
338
                       (uint8_t *)left, stride, log2_size, c_idx);
339
5402678
        break;
340
27029988
    default:
341
27029988
        s->hpc.pred_angular[log2_size - 2]((uint8_t *)src, (uint8_t *)top,
342
                                           (uint8_t *)left, stride, c_idx,
343
                                           mode);
344
27029988
        break;
345
    }
346
}
347
348
#define INTRA_PRED(size)                                                            \
349
static void FUNC(intra_pred_ ## size)(HEVCContext *s, int x0, int y0, int c_idx)    \
350
{                                                                                   \
351
    FUNC(intra_pred)(s, x0, y0, size, c_idx);                                       \
352
}
353
354
29211212
INTRA_PRED(2)
355
4274090
INTRA_PRED(3)
356
1427045
INTRA_PRED(4)
357
281682
INTRA_PRED(5)
358
359
#undef INTRA_PRED
360
361
8744180
static av_always_inline void FUNC(pred_planar)(uint8_t *_src, const uint8_t *_top,
362
                                  const uint8_t *_left, ptrdiff_t stride,
363
                                  int trafo_size)
364
{
365
    int x, y;
366
8744180
    pixel *src        = (pixel *)_src;
367
8744180
    const pixel *top  = (const pixel *)_top;
368
8744180
    const pixel *left = (const pixel *)_left;
369
8744180
    int size = 1 << trafo_size;
370
64393164
    for (y = 0; y < size; y++)
371
619046328
        for (x = 0; x < size; x++)
372
563397344
            POS(x, y) = ((size - 1 - x) * left[y] + (x + 1) * top[size]  +
373
563397344
                         (size - 1 - y) * top[x]  + (y + 1) * left[size] + size) >> (trafo_size + 1);
374
}
375
376
#define PRED_PLANAR(size)\
377
static void FUNC(pred_planar_ ## size)(uint8_t *src, const uint8_t *top,        \
378
                                       const uint8_t *left, ptrdiff_t stride)   \
379
{                                                                               \
380
    FUNC(pred_planar)(src, top, left, stride, size + 2);                        \
381
}
382
383
5969454
PRED_PLANAR(0)
384
946692
PRED_PLANAR(1)
385
361839
PRED_PLANAR(2)
386
78832
PRED_PLANAR(3)
387
388
#undef PRED_PLANAR
389
390
5402678
static void FUNC(pred_dc)(uint8_t *_src, const uint8_t *_top,
391
                          const uint8_t *_left,
392
                          ptrdiff_t stride, int log2_size, int c_idx)
393
{
394
    int i, j, x, y;
395
5402678
    int size          = (1 << log2_size);
396
5402678
    pixel *src        = (pixel *)_src;
397
5402678
    const pixel *top  = (const pixel *)_top;
398
5402678
    const pixel *left = (const pixel *)_left;
399
5402678
    int dc            = size;
400
    pixel4 a;
401
43196214
    for (i = 0; i < size; i++)
402
37793536
        dc += left[i] + top[i];
403
404
5402678
    dc >>= log2_size + 1;
405
406
5402678
    a = PIXEL_SPLAT_X4(dc);
407
408
43196214
    for (i = 0; i < size; i++)
409
145802640
        for (j = 0; j < size; j+=4)
410
108009104
            AV_WN4P(&POS(j, i), a);
411
412

5402678
    if (c_idx == 0 && size < 32) {
413
1885428
        POS(0, 0) = (left[0] + 2 * dc + top[0] + 2) >> 2;
414
12560880
        for (x = 1; x < size; x++)
415
10675452
            POS(x, 0) = (top[x] + 3 * dc + 2) >> 2;
416
12560880
        for (y = 1; y < size; y++)
417
10675452
            POS(0, y) = (left[y] + 3 * dc + 2) >> 2;
418
    }
419
}
420
421
27029988
static av_always_inline void FUNC(pred_angular)(uint8_t *_src,
422
                                                const uint8_t *_top,
423
                                                const uint8_t *_left,
424
                                                ptrdiff_t stride, int c_idx,
425
                                                int mode, int size)
426
{
427
    int x, y;
428
27029988
    pixel *src        = (pixel *)_src;
429
27029988
    const pixel *top  = (const pixel *)_top;
430
27029988
    const pixel *left = (const pixel *)_left;
431
432
    static const int intra_pred_angle[] = {
433
         32,  26,  21,  17, 13,  9,  5, 2, 0, -2, -5, -9, -13, -17, -21, -26, -32,
434
        -26, -21, -17, -13, -9, -5, -2, 0, 2,  5,  9, 13,  17,  21,  26,  32
435
    };
436
    static const int inv_angle[] = {
437
        -4096, -1638, -910, -630, -482, -390, -315, -256, -315, -390, -482,
438
        -630, -910, -1638, -4096
439
    };
440
441
27029988
    int angle = intra_pred_angle[mode - 2];
442
    pixel ref_array[3 * MAX_TB_SIZE + 4];
443
27029988
    pixel *ref_tmp = ref_array + size;
444
    const pixel *ref;
445
27029988
    int last = (size * angle) >> 5;
446
447
27029988
    if (mode >= 18) {
448
14665852
        ref = top - 1;
449

14665852
        if (angle < 0 && last < -1) {
450
12518796
            for (x = 0; x <= size; x += 4)
451
8912164
                AV_WN4P(&ref_tmp[x], AV_RN4P(&top[x - 1]));
452
16773316
            for (x = last; x <= -1; x++)
453
13166684
                ref_tmp[x] = left[-1 + ((x * inv_angle[mode - 11] + 128) >> 8)];
454
3606632
            ref = ref_tmp;
455
        }
456
457
98257948
        for (y = 0; y < size; y++) {
458
83592096
            int idx  = ((y + 1) * angle) >> 5;
459
83592096
            int fact = ((y + 1) * angle) & 31;
460
83592096
            if (fact) {
461
152640478
                for (x = 0; x < size; x += 4) {
462
102423048
                    POS(x    , y) = ((32 - fact) * ref[x + idx + 1] +
463
102423048
                                           fact  * ref[x + idx + 2] + 16) >> 5;
464
102423048
                    POS(x + 1, y) = ((32 - fact) * ref[x + 1 + idx + 1] +
465
102423048
                                           fact  * ref[x + 1 + idx + 2] + 16) >> 5;
466
102423048
                    POS(x + 2, y) = ((32 - fact) * ref[x + 2 + idx + 1] +
467
102423048
                                           fact  * ref[x + 2 + idx + 2] + 16) >> 5;
468
102423048
                    POS(x + 3, y) = ((32 - fact) * ref[x + 3 + idx + 1] +
469
102423048
                                           fact  * ref[x + 3 + idx + 2] + 16) >> 5;
470
                }
471
            } else {
472
105896562
                for (x = 0; x < size; x += 4)
473
72521896
                    AV_WN4P(&POS(x, y), AV_RN4P(&ref[x + idx + 1]));
474
            }
475
        }
476

14665852
        if (mode == 26 && c_idx == 0 && size < 32) {
477
11493756
            for (y = 0; y < size; y++)
478
9844904
                POS(0, y) = av_clip_pixel(top[0] + ((left[y] - left[-1]) >> 1));
479
        }
480
    } else {
481
12364136
        ref = left - 1;
482

12364136
        if (angle < 0 && last < -1) {
483
8480438
            for (x = 0; x <= size; x += 4)
484
6063152
                AV_WN4P(&ref_tmp[x], AV_RN4P(&left[x - 1]));
485
10601312
            for (x = last; x <= -1; x++)
486
8184026
                ref_tmp[x] = top[-1 + ((x * inv_angle[mode - 11] + 128) >> 8)];
487
2417286
            ref = ref_tmp;
488
        }
489
490
84252896
        for (x = 0; x < size; x++) {
491
71888760
            int idx  = ((x + 1) * angle) >> 5;
492
71888760
            int fact = ((x + 1) * angle) & 31;
493
71888760
            if (fact) {
494
497748986
                for (y = 0; y < size; y++) {
495
446049280
                    POS(x, y) = ((32 - fact) * ref[y + idx + 1] +
496
446049280
                                       fact  * ref[y + idx + 2] + 16) >> 5;
497
                }
498
            } else {
499
200920926
                for (y = 0; y < size; y++)
500
180731872
                    POS(x, y) = ref[y + idx + 1];
501
            }
502
        }
503

12364136
        if (mode == 10 && c_idx == 0 && size < 32) {
504
2111276
            for (x = 0; x < size; x += 4) {
505
1257248
                POS(x,     0) = av_clip_pixel(left[0] + ((top[x    ] - top[-1]) >> 1));
506
1257248
                POS(x + 1, 0) = av_clip_pixel(left[0] + ((top[x + 1] - top[-1]) >> 1));
507
1257248
                POS(x + 2, 0) = av_clip_pixel(left[0] + ((top[x + 2] - top[-1]) >> 1));
508
1257248
                POS(x + 3, 0) = av_clip_pixel(left[0] + ((top[x + 3] - top[-1]) >> 1));
509
            }
510
        }
511
    }
512
}
513
514
19839538
static void FUNC(pred_angular_0)(uint8_t *src, const uint8_t *top,
515
                                 const uint8_t *left,
516
                                 ptrdiff_t stride, int c_idx, int mode)
517
{
518
19839538
    FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 2);
519
}
520
521
5399734
static void FUNC(pred_angular_1)(uint8_t *src, const uint8_t *top,
522
                                 const uint8_t *left,
523
                                 ptrdiff_t stride, int c_idx, int mode)
524
{
525
5399734
    FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 3);
526
}
527
528
1523630
static void FUNC(pred_angular_2)(uint8_t *src, const uint8_t *top,
529
                                 const uint8_t *left,
530
                                 ptrdiff_t stride, int c_idx, int mode)
531
{
532
1523630
    FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 4);
533
}
534
535
267086
static void FUNC(pred_angular_3)(uint8_t *src, const uint8_t *top,
536
                                 const uint8_t *left,
537
                                 ptrdiff_t stride, int c_idx, int mode)
538
{
539
267086
    FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 5);
540
}
541
542
#undef EXTEND_LEFT_CIP
543
#undef EXTEND_RIGHT_CIP
544
#undef EXTEND_UP_CIP
545
#undef EXTEND_DOWN_CIP
546
#undef IS_INTRA
547
#undef MVF_PU
548
#undef MVF
549
#undef PU
550
#undef EXTEND
551
#undef MIN_TB_ADDR_ZS
552
#undef POS