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
|
/*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#include <assert.h>
#include <string.h>
#include "config/aom_config.h"
#include "config/aom_dsp_rtcd.h"
#include "aom/aom_integer.h"
#include "aom_dsp/aom_convolve.h"
#include "aom_dsp/aom_dsp_common.h"
#include "aom_dsp/aom_filter.h"
#include "aom_ports/mem.h"
static INLINE int horz_scalar_product(const uint8_t *a, const int16_t *b) {
int sum = 0;
for (int k = 0; k < SUBPEL_TAPS; ++k) sum += a[k] * b[k];
return sum;
}
static INLINE int vert_scalar_product(const uint8_t *a, ptrdiff_t a_stride,
const int16_t *b) {
int sum = 0;
for (int k = 0; k < SUBPEL_TAPS; ++k) sum += a[k * a_stride] * b[k];
return sum;
}
static void convolve_horiz(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *x_filters, int x0_q4,
int x_step_q4, int w, int h) {
src -= SUBPEL_TAPS / 2 - 1;
for (int y = 0; y < h; ++y) {
int x_q4 = x0_q4;
for (int x = 0; x < w; ++x) {
const uint8_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK];
const int sum = horz_scalar_product(src_x, x_filter);
dst[x] = clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS));
x_q4 += x_step_q4;
}
src += src_stride;
dst += dst_stride;
}
}
static void convolve_vert(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *y_filters, int y0_q4,
int y_step_q4, int w, int h) {
src -= src_stride * (SUBPEL_TAPS / 2 - 1);
for (int x = 0; x < w; ++x) {
int y_q4 = y0_q4;
for (int y = 0; y < h; ++y) {
const unsigned char *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride];
const int16_t *const y_filter = y_filters[y_q4 & SUBPEL_MASK];
const int sum = vert_scalar_product(src_y, src_stride, y_filter);
dst[y * dst_stride] = clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS));
y_q4 += y_step_q4;
}
++src;
++dst;
}
}
static const InterpKernel *get_filter_base(const int16_t *filter) {
// NOTE: This assumes that the filter table is 256-byte aligned.
// TODO(agrange) Modify to make independent of table alignment.
return (const InterpKernel *)(((intptr_t)filter) & ~((intptr_t)0xFF));
}
static int get_filter_offset(const int16_t *f, const InterpKernel *base) {
return (int)((const InterpKernel *)(intptr_t)f - base);
}
void aom_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4, int w,
int h) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
(void)filter_y;
(void)y_step_q4;
convolve_horiz(src, src_stride, dst, dst_stride, filters_x, x0_q4, x_step_q4,
w, h);
}
void aom_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4, int w,
int h) {
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
(void)filter_x;
(void)x_step_q4;
convolve_vert(src, src_stride, dst, dst_stride, filters_y, y0_q4, y_step_q4,
w, h);
}
void aom_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst,
ptrdiff_t dst_stride, const int16_t *filter_x,
int filter_x_stride, const int16_t *filter_y,
int filter_y_stride, int w, int h) {
int r;
(void)filter_x;
(void)filter_x_stride;
(void)filter_y;
(void)filter_y_stride;
for (r = h; r > 0; --r) {
memcpy(dst, src, w);
src += src_stride;
dst += dst_stride;
}
}
static INLINE int highbd_vert_scalar_product(const uint16_t *a,
ptrdiff_t a_stride,
const int16_t *b) {
int sum = 0;
for (int k = 0; k < SUBPEL_TAPS; ++k) sum += a[k * a_stride] * b[k];
return sum;
}
static INLINE int highbd_horz_scalar_product(const uint16_t *a,
const int16_t *b) {
int sum = 0;
for (int k = 0; k < SUBPEL_TAPS; ++k) sum += a[k] * b[k];
return sum;
}
static void highbd_convolve_horiz(const uint8_t *src8, ptrdiff_t src_stride,
uint8_t *dst8, ptrdiff_t dst_stride,
const InterpKernel *x_filters, int x0_q4,
int x_step_q4, int w, int h, int bd) {
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
src -= SUBPEL_TAPS / 2 - 1;
for (int y = 0; y < h; ++y) {
int x_q4 = x0_q4;
for (int x = 0; x < w; ++x) {
const uint16_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK];
const int sum = highbd_horz_scalar_product(src_x, x_filter);
dst[x] = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
x_q4 += x_step_q4;
}
src += src_stride;
dst += dst_stride;
}
}
static void highbd_convolve_vert(const uint8_t *src8, ptrdiff_t src_stride,
uint8_t *dst8, ptrdiff_t dst_stride,
const InterpKernel *y_filters, int y0_q4,
int y_step_q4, int w, int h, int bd) {
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
src -= src_stride * (SUBPEL_TAPS / 2 - 1);
for (int x = 0; x < w; ++x) {
int y_q4 = y0_q4;
for (int y = 0; y < h; ++y) {
const uint16_t *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride];
const int16_t *const y_filter = y_filters[y_q4 & SUBPEL_MASK];
const int sum = highbd_vert_scalar_product(src_y, src_stride, y_filter);
dst[y * dst_stride] =
clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
y_q4 += y_step_q4;
}
++src;
++dst;
}
}
void aom_highbd_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4, int w,
int h, int bd) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
(void)filter_y;
(void)y_step_q4;
highbd_convolve_horiz(src, src_stride, dst, dst_stride, filters_x, x0_q4,
x_step_q4, w, h, bd);
}
void aom_highbd_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4, int w,
int h, int bd) {
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
(void)filter_x;
(void)x_step_q4;
highbd_convolve_vert(src, src_stride, dst, dst_stride, filters_y, y0_q4,
y_step_q4, w, h, bd);
}
void aom_highbd_convolve_copy_c(const uint8_t *src8, ptrdiff_t src_stride,
uint8_t *dst8, ptrdiff_t dst_stride,
const int16_t *filter_x, int filter_x_stride,
const int16_t *filter_y, int filter_y_stride,
int w, int h, int bd) {
int r;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
(void)filter_x;
(void)filter_y;
(void)filter_x_stride;
(void)filter_y_stride;
(void)bd;
for (r = h; r > 0; --r) {
memcpy(dst, src, w * sizeof(uint16_t));
src += src_stride;
dst += dst_stride;
}
}
|
