summaryrefslogtreecommitdiffstats
path: root/media/libvpx/vpx_dsp/sad.c
blob: c0c3ff99645cb80752f746c54460ea991340fab8 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
/*
 *  Copyright (c) 2015 The WebM project authors. All Rights Reserved.
 *
 *  Use of this source code is governed by a BSD-style license
 *  that can be found in the LICENSE file in the root of the source
 *  tree. An additional intellectual property rights grant can be found
 *  in the file PATENTS.  All contributing project authors may
 *  be found in the AUTHORS file in the root of the source tree.
 */

#include <stdlib.h>

#include "./vpx_config.h"
#include "./vpx_dsp_rtcd.h"

#include "vpx/vpx_integer.h"
#include "vpx_ports/mem.h"

/* Sum the difference between every corresponding element of the buffers. */
static INLINE unsigned int sad(const uint8_t *a, int a_stride,
                               const uint8_t *b, int b_stride,
                               int width, int height) {
  int y, x;
  unsigned int sad = 0;

  for (y = 0; y < height; y++) {
    for (x = 0; x < width; x++)
      sad += abs(a[x] - b[x]);

    a += a_stride;
    b += b_stride;
  }
  return sad;
}

// TODO(johannkoenig): this moved to vpx_dsp, should be able to clean this up.
/* Remove dependency on vp9 variance function by duplicating vp9_comp_avg_pred.
 * The function averages every corresponding element of the buffers and stores
 * the value in a third buffer, comp_pred.
 * pred and comp_pred are assumed to have stride = width
 * In the usage below comp_pred is a local array.
 */
static INLINE void avg_pred(uint8_t *comp_pred, const uint8_t *pred, int width,
                            int height, const uint8_t *ref, int ref_stride) {
  int i, j;

  for (i = 0; i < height; i++) {
    for (j = 0; j < width; j++) {
      const int tmp = pred[j] + ref[j];
      comp_pred[j] = ROUND_POWER_OF_TWO(tmp, 1);
    }
    comp_pred += width;
    pred += width;
    ref += ref_stride;
  }
}

#if CONFIG_VP9_HIGHBITDEPTH
static INLINE void highbd_avg_pred(uint16_t *comp_pred, const uint8_t *pred8,
                                   int width, int height, const uint8_t *ref8,
                                   int ref_stride) {
  int i, j;
  uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
  uint16_t *ref = CONVERT_TO_SHORTPTR(ref8);
  for (i = 0; i < height; i++) {
    for (j = 0; j < width; j++) {
      const int tmp = pred[j] + ref[j];
      comp_pred[j] = ROUND_POWER_OF_TWO(tmp, 1);
    }
    comp_pred += width;
    pred += width;
    ref += ref_stride;
  }
}
#endif  // CONFIG_VP9_HIGHBITDEPTH

#define sadMxN(m, n) \
unsigned int vpx_sad##m##x##n##_c(const uint8_t *src, int src_stride, \
                                  const uint8_t *ref, int ref_stride) { \
  return sad(src, src_stride, ref, ref_stride, m, n); \
} \
unsigned int vpx_sad##m##x##n##_avg_c(const uint8_t *src, int src_stride, \
                                      const uint8_t *ref, int ref_stride, \
                                      const uint8_t *second_pred) { \
  uint8_t comp_pred[m * n]; \
  avg_pred(comp_pred, second_pred, m, n, ref, ref_stride); \
  return sad(src, src_stride, comp_pred, m, m, n); \
}

// depending on call sites, pass **ref_array to avoid & in subsequent call and
// de-dup with 4D below.
#define sadMxNxK(m, n, k) \
void vpx_sad##m##x##n##x##k##_c(const uint8_t *src, int src_stride, \
                                const uint8_t *ref_array, int ref_stride, \
                                uint32_t *sad_array) { \
  int i; \
  for (i = 0; i < k; ++i) \
    sad_array[i] = vpx_sad##m##x##n##_c(src, src_stride, &ref_array[i], ref_stride); \
}

// This appears to be equivalent to the above when k == 4 and refs is const
#define sadMxNx4D(m, n) \
void vpx_sad##m##x##n##x4d_c(const uint8_t *src, int src_stride, \
                             const uint8_t *const ref_array[], int ref_stride, \
                             uint32_t *sad_array) { \
  int i; \
  for (i = 0; i < 4; ++i) \
    sad_array[i] = vpx_sad##m##x##n##_c(src, src_stride, ref_array[i], ref_stride); \
}

// 64x64
sadMxN(64, 64)
sadMxNxK(64, 64, 3)
sadMxNxK(64, 64, 8)
sadMxNx4D(64, 64)

// 64x32
sadMxN(64, 32)
sadMxNx4D(64, 32)

// 32x64
sadMxN(32, 64)
sadMxNx4D(32, 64)

// 32x32
sadMxN(32, 32)
sadMxNxK(32, 32, 3)
sadMxNxK(32, 32, 8)
sadMxNx4D(32, 32)

// 32x16
sadMxN(32, 16)
sadMxNx4D(32, 16)

// 16x32
sadMxN(16, 32)
sadMxNx4D(16, 32)

// 16x16
sadMxN(16, 16)
sadMxNxK(16, 16, 3)
sadMxNxK(16, 16, 8)
sadMxNx4D(16, 16)

// 16x8
sadMxN(16, 8)
sadMxNxK(16, 8, 3)
sadMxNxK(16, 8, 8)
sadMxNx4D(16, 8)

// 8x16
sadMxN(8, 16)
sadMxNxK(8, 16, 3)
sadMxNxK(8, 16, 8)
sadMxNx4D(8, 16)

// 8x8
sadMxN(8, 8)
sadMxNxK(8, 8, 3)
sadMxNxK(8, 8, 8)
sadMxNx4D(8, 8)

// 8x4
sadMxN(8, 4)
sadMxNxK(8, 4, 8)
sadMxNx4D(8, 4)

// 4x8
sadMxN(4, 8)
sadMxNxK(4, 8, 8)
sadMxNx4D(4, 8)

// 4x4
sadMxN(4, 4)
sadMxNxK(4, 4, 3)
sadMxNxK(4, 4, 8)
sadMxNx4D(4, 4)

#if CONFIG_VP9_HIGHBITDEPTH
static INLINE unsigned int highbd_sad(const uint8_t *a8, int a_stride,
                                      const uint8_t *b8, int b_stride,
                                      int width, int height) {
  int y, x;
  unsigned int sad = 0;
  const uint16_t *a = CONVERT_TO_SHORTPTR(a8);
  const uint16_t *b = CONVERT_TO_SHORTPTR(b8);
  for (y = 0; y < height; y++) {
    for (x = 0; x < width; x++)
      sad += abs(a[x] - b[x]);

    a += a_stride;
    b += b_stride;
  }
  return sad;
}

static INLINE unsigned int highbd_sadb(const uint8_t *a8, int a_stride,
                                       const uint16_t *b, int b_stride,
                                       int width, int height) {
  int y, x;
  unsigned int sad = 0;
  const uint16_t *a = CONVERT_TO_SHORTPTR(a8);
  for (y = 0; y < height; y++) {
    for (x = 0; x < width; x++)
      sad += abs(a[x] - b[x]);

    a += a_stride;
    b += b_stride;
  }
  return sad;
}

#define highbd_sadMxN(m, n) \
unsigned int vpx_highbd_sad##m##x##n##_c(const uint8_t *src, int src_stride, \
                                         const uint8_t *ref, int ref_stride) { \
  return highbd_sad(src, src_stride, ref, ref_stride, m, n); \
} \
unsigned int vpx_highbd_sad##m##x##n##_avg_c(const uint8_t *src, \
                                             int src_stride, \
                                             const uint8_t *ref, \
                                             int ref_stride, \
                                             const uint8_t *second_pred) { \
  uint16_t comp_pred[m * n]; \
  highbd_avg_pred(comp_pred, second_pred, m, n, ref, ref_stride); \
  return highbd_sadb(src, src_stride, comp_pred, m, m, n); \
}

#define highbd_sadMxNxK(m, n, k) \
void vpx_highbd_sad##m##x##n##x##k##_c(const uint8_t *src, int src_stride, \
                                       const uint8_t *ref_array, int ref_stride, \
                                       uint32_t *sad_array) { \
  int i; \
  for (i = 0; i < k; ++i) { \
    sad_array[i] = vpx_highbd_sad##m##x##n##_c(src, src_stride, &ref_array[i], \
                                               ref_stride); \
  } \
}

#define highbd_sadMxNx4D(m, n) \
void vpx_highbd_sad##m##x##n##x4d_c(const uint8_t *src, int src_stride, \
                                    const uint8_t *const ref_array[], \
                                    int ref_stride, uint32_t *sad_array) { \
  int i; \
  for (i = 0; i < 4; ++i) { \
    sad_array[i] = vpx_highbd_sad##m##x##n##_c(src, src_stride, ref_array[i], \
                                               ref_stride); \
  } \
}

// 64x64
highbd_sadMxN(64, 64)
highbd_sadMxNxK(64, 64, 3)
highbd_sadMxNxK(64, 64, 8)
highbd_sadMxNx4D(64, 64)

// 64x32
highbd_sadMxN(64, 32)
highbd_sadMxNx4D(64, 32)

// 32x64
highbd_sadMxN(32, 64)
highbd_sadMxNx4D(32, 64)

// 32x32
highbd_sadMxN(32, 32)
highbd_sadMxNxK(32, 32, 3)
highbd_sadMxNxK(32, 32, 8)
highbd_sadMxNx4D(32, 32)

// 32x16
highbd_sadMxN(32, 16)
highbd_sadMxNx4D(32, 16)

// 16x32
highbd_sadMxN(16, 32)
highbd_sadMxNx4D(16, 32)

// 16x16
highbd_sadMxN(16, 16)
highbd_sadMxNxK(16, 16, 3)
highbd_sadMxNxK(16, 16, 8)
highbd_sadMxNx4D(16, 16)

// 16x8
highbd_sadMxN(16, 8)
highbd_sadMxNxK(16, 8, 3)
highbd_sadMxNxK(16, 8, 8)
highbd_sadMxNx4D(16, 8)

// 8x16
highbd_sadMxN(8, 16)
highbd_sadMxNxK(8, 16, 3)
highbd_sadMxNxK(8, 16, 8)
highbd_sadMxNx4D(8, 16)

// 8x8
highbd_sadMxN(8, 8)
highbd_sadMxNxK(8, 8, 3)
highbd_sadMxNxK(8, 8, 8)
highbd_sadMxNx4D(8, 8)

// 8x4
highbd_sadMxN(8, 4)
highbd_sadMxNxK(8, 4, 8)
highbd_sadMxNx4D(8, 4)

// 4x8
highbd_sadMxN(4, 8)
highbd_sadMxNxK(4, 8, 8)
highbd_sadMxNx4D(4, 8)

// 4x4
highbd_sadMxN(4, 4)
highbd_sadMxNxK(4, 4, 3)
highbd_sadMxNxK(4, 4, 8)
highbd_sadMxNx4D(4, 4)

#endif  // CONFIG_VP9_HIGHBITDEPTH