FFmpeg coverage

Directory: ../../../ffmpeg/
File: src/libavcodec/x86/vp8dsp_init.c
Date: 2026-07-18 19:31:15
Exec Total Coverage
Lines: 80 80 100.0%
Functions: 29 29 100.0%
Branches: 18 18 100.0%
Line Branch Exec Source
1 /*
2 * VP8 DSP functions x86-optimized
3 * Copyright (c) 2010 Ronald S. Bultje <rsbultje@gmail.com>
4 * Copyright (c) 2010 Fiona Glaser <fiona@x264.com>
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/attributes.h"
24 #include "libavutil/cpu.h"
25 #include "libavutil/mem_internal.h"
26 #include "libavutil/x86/cpu.h"
27 #include "libavcodec/vp8dsp.h"
28
29 /*
30 * MC functions
31 */
32 void ff_put_vp8_epel8_h4_sse2 (uint8_t *dst, ptrdiff_t dststride,
33 const uint8_t *src, ptrdiff_t srcstride,
34 int height, int mx, int my);
35 void ff_put_vp8_epel8_h6_sse2 (uint8_t *dst, ptrdiff_t dststride,
36 const uint8_t *src, ptrdiff_t srcstride,
37 int height, int mx, int my);
38 void ff_put_vp8_epel8_v4_sse2 (uint8_t *dst, ptrdiff_t dststride,
39 const uint8_t *src, ptrdiff_t srcstride,
40 int height, int mx, int my);
41 void ff_put_vp8_epel8_v6_sse2 (uint8_t *dst, ptrdiff_t dststride,
42 const uint8_t *src, ptrdiff_t srcstride,
43 int height, int mx, int my);
44
45 void ff_put_vp8_epel4_h4_ssse3 (uint8_t *dst, ptrdiff_t dststride,
46 const uint8_t *src, ptrdiff_t srcstride,
47 int height, int mx, int my);
48 void ff_put_vp8_epel4_h6_ssse3 (uint8_t *dst, ptrdiff_t dststride,
49 const uint8_t *src, ptrdiff_t srcstride,
50 int height, int mx, int my);
51 void ff_put_vp8_epel4_v4_ssse3 (uint8_t *dst, ptrdiff_t dststride,
52 const uint8_t *src, ptrdiff_t srcstride,
53 int height, int mx, int my);
54 void ff_put_vp8_epel4_v6_ssse3 (uint8_t *dst, ptrdiff_t dststride,
55 const uint8_t *src, ptrdiff_t srcstride,
56 int height, int mx, int my);
57 void ff_put_vp8_epel8_h4_ssse3 (uint8_t *dst, ptrdiff_t dststride,
58 const uint8_t *src, ptrdiff_t srcstride,
59 int height, int mx, int my);
60 void ff_put_vp8_epel8_h6_ssse3 (uint8_t *dst, ptrdiff_t dststride,
61 const uint8_t *src, ptrdiff_t srcstride,
62 int height, int mx, int my);
63 void ff_put_vp8_epel8_v4_ssse3 (uint8_t *dst, ptrdiff_t dststride,
64 const uint8_t *src, ptrdiff_t srcstride,
65 int height, int mx, int my);
66 void ff_put_vp8_epel8_v6_ssse3 (uint8_t *dst, ptrdiff_t dststride,
67 const uint8_t *src, ptrdiff_t srcstride,
68 int height, int mx, int my);
69
70 void ff_put_vp8_bilinear8_h_sse2 (uint8_t *dst, ptrdiff_t dststride,
71 const uint8_t *src, ptrdiff_t srcstride,
72 int height, int mx, int my);
73 void ff_put_vp8_bilinear4_h_ssse3 (uint8_t *dst, ptrdiff_t dststride,
74 const uint8_t *src, ptrdiff_t srcstride,
75 int height, int mx, int my);
76 void ff_put_vp8_bilinear8_h_ssse3 (uint8_t *dst, ptrdiff_t dststride,
77 const uint8_t *src, ptrdiff_t srcstride,
78 int height, int mx, int my);
79
80 void ff_put_vp8_bilinear8_v_sse2 (uint8_t *dst, ptrdiff_t dststride,
81 const uint8_t *src, ptrdiff_t srcstride,
82 int height, int mx, int my);
83 void ff_put_vp8_bilinear4_v_ssse3 (uint8_t *dst, ptrdiff_t dststride,
84 const uint8_t *src, ptrdiff_t srcstride,
85 int height, int mx, int my);
86 void ff_put_vp8_bilinear8_v_ssse3 (uint8_t *dst, ptrdiff_t dststride,
87 const uint8_t *src, ptrdiff_t srcstride,
88 int height, int mx, int my);
89
90
91 void ff_put_vp8_pixels8_sse2(uint8_t *dst, ptrdiff_t dststride,
92 const uint8_t *src, ptrdiff_t srcstride,
93 int height, int mx, int my);
94 void ff_put_vp8_pixels16_sse(uint8_t *dst, ptrdiff_t dststride,
95 const uint8_t *src, ptrdiff_t srcstride,
96 int height, int mx, int my);
97
98 #define TAP_W16(OPT, FILTERTYPE, TAPTYPE) \
99 static void ff_put_vp8_ ## FILTERTYPE ## 16_ ## TAPTYPE ## _ ## OPT( \
100 uint8_t *dst, ptrdiff_t dststride, const uint8_t *src, \
101 ptrdiff_t srcstride, int height, int mx, int my) \
102 { \
103 ff_put_vp8_ ## FILTERTYPE ## 8_ ## TAPTYPE ## _ ## OPT( \
104 dst, dststride, src, srcstride, height, mx, my); \
105 ff_put_vp8_ ## FILTERTYPE ## 8_ ## TAPTYPE ## _ ## OPT( \
106 dst + 8, dststride, src + 8, srcstride, height, mx, my); \
107 }
108
109 6 TAP_W16(sse2, epel, h6)
110 6 TAP_W16(sse2, epel, v6)
111 6 TAP_W16(sse2, bilinear, h)
112 6 TAP_W16(sse2, bilinear, v)
113
114 4 TAP_W16(ssse3, epel, h6)
115 4 TAP_W16(ssse3, epel, v6)
116 4 TAP_W16(ssse3, bilinear, h)
117 4 TAP_W16(ssse3, bilinear, v)
118
119 #define HVTAP(OPT, ALIGN, TAPNUMX, TAPNUMY, SIZE, MAXHEIGHT) \
120 static void ff_put_vp8_epel ## SIZE ## _h ## TAPNUMX ## v ## TAPNUMY ## _ ## OPT( \
121 uint8_t *dst, ptrdiff_t dststride, const uint8_t *src, \
122 ptrdiff_t srcstride, int height, int mx, int my) \
123 { \
124 LOCAL_ALIGNED(ALIGN, uint8_t, tmp, [SIZE * (MAXHEIGHT + TAPNUMY - 1)]); \
125 uint8_t *tmpptr = tmp + SIZE * (TAPNUMY / 2 - 1); \
126 src -= srcstride * (TAPNUMY / 2 - 1); \
127 ff_put_vp8_epel ## SIZE ## _h ## TAPNUMX ## _ ## OPT( \
128 tmp, SIZE, src, srcstride, height + TAPNUMY - 1, mx, my); \
129 ff_put_vp8_epel ## SIZE ## _v ## TAPNUMY ## _ ## OPT( \
130 dst, dststride, tmpptr, SIZE, height, mx, my); \
131 }
132
133 #define HVTAPSSE2(x, y, w) \
134 HVTAP(sse2, 16, x, y, w, 16) \
135 HVTAP(ssse3, 16, x, y, w, 16)
136
137 10 HVTAPSSE2(4, 4, 8)
138 10 HVTAPSSE2(4, 6, 8)
139 10 HVTAPSSE2(6, 4, 8)
140 10 HVTAPSSE2(6, 6, 8)
141 10 HVTAPSSE2(6, 6, 16)
142
143 2 HVTAP(ssse3, 16, 4, 4, 4, 8)
144 2 HVTAP(ssse3, 16, 4, 6, 4, 8)
145 2 HVTAP(ssse3, 16, 6, 4, 4, 8)
146 2 HVTAP(ssse3, 16, 6, 6, 4, 8)
147
148 #define HVBILIN(OPT, ALIGN, SIZE, MAXHEIGHT) \
149 static void ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT( \
150 uint8_t *dst, ptrdiff_t dststride, const uint8_t *src, \
151 ptrdiff_t srcstride, int height, int mx, int my) \
152 { \
153 LOCAL_ALIGNED(ALIGN, uint8_t, tmp, [SIZE * (MAXHEIGHT + 2)]); \
154 ff_put_vp8_bilinear ## SIZE ## _h_ ## OPT( \
155 tmp, SIZE, src, srcstride, height + 1, mx, my); \
156 ff_put_vp8_bilinear ## SIZE ## _v_ ## OPT( \
157 dst, dststride, tmp, SIZE, height, mx, my); \
158 }
159
160 3 HVBILIN(sse2, 8, 8, 16)
161 3 HVBILIN(sse2, 8, 16, 16)
162 2 HVBILIN(ssse3, 8, 4, 8)
163 2 HVBILIN(ssse3, 8, 8, 16)
164 2 HVBILIN(ssse3, 8, 16, 16)
165
166 void ff_vp8_idct_dc_add_sse2(uint8_t *dst, int16_t block[16],
167 ptrdiff_t stride);
168 void ff_vp8_idct_dc_add_sse4(uint8_t *dst, int16_t block[16],
169 ptrdiff_t stride);
170 void ff_vp8_idct_dc_add4y_sse2(uint8_t *dst, int16_t block[4][16],
171 ptrdiff_t stride);
172 void ff_vp8_idct_dc_add4uv_mmx(uint8_t *dst, int16_t block[2][16],
173 ptrdiff_t stride);
174 void ff_vp8_luma_dc_wht_sse(int16_t block[4][4][16], int16_t dc[16]);
175 void ff_vp8_idct_add_sse(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
176
177 #define DECLARE_LOOP_FILTER(NAME) \
178 void ff_vp8_v_loop_filter_simple_ ## NAME(uint8_t *dst, \
179 ptrdiff_t stride, \
180 int flim); \
181 void ff_vp8_h_loop_filter_simple_ ## NAME(uint8_t *dst, \
182 ptrdiff_t stride, \
183 int flim); \
184 void ff_vp8_v_loop_filter16y_inner_ ## NAME (uint8_t *dst, \
185 ptrdiff_t stride, \
186 int e, int i, int hvt); \
187 void ff_vp8_h_loop_filter16y_inner_ ## NAME (uint8_t *dst, \
188 ptrdiff_t stride, \
189 int e, int i, int hvt); \
190 void ff_vp8_v_loop_filter8uv_inner_ ## NAME (uint8_t *dstU, \
191 uint8_t *dstV, \
192 ptrdiff_t s, \
193 int e, int i, int hvt); \
194 void ff_vp8_h_loop_filter8uv_inner_ ## NAME (uint8_t *dstU, \
195 uint8_t *dstV, \
196 ptrdiff_t s, \
197 int e, int i, int hvt); \
198 void ff_vp8_v_loop_filter16y_mbedge_ ## NAME(uint8_t *dst, \
199 ptrdiff_t stride, \
200 int e, int i, int hvt); \
201 void ff_vp8_h_loop_filter16y_mbedge_ ## NAME(uint8_t *dst, \
202 ptrdiff_t stride, \
203 int e, int i, int hvt); \
204 void ff_vp8_v_loop_filter8uv_mbedge_ ## NAME(uint8_t *dstU, \
205 uint8_t *dstV, \
206 ptrdiff_t s, \
207 int e, int i, int hvt); \
208 void ff_vp8_h_loop_filter8uv_mbedge_ ## NAME(uint8_t *dstU, \
209 uint8_t *dstV, \
210 ptrdiff_t s, \
211 int e, int i, int hvt);
212
213 DECLARE_LOOP_FILTER(sse2)
214 DECLARE_LOOP_FILTER(ssse3)
215 DECLARE_LOOP_FILTER(sse4)
216
217 #define VP8_LUMA_MC_FUNC(IDX, SIZE, OPT) \
218 c->put_vp8_epel_pixels_tab[IDX][0][2] = ff_put_vp8_epel ## SIZE ## _h6_ ## OPT; \
219 c->put_vp8_epel_pixels_tab[IDX][2][0] = ff_put_vp8_epel ## SIZE ## _v6_ ## OPT; \
220 c->put_vp8_epel_pixels_tab[IDX][2][2] = ff_put_vp8_epel ## SIZE ## _h6v6_ ## OPT
221
222 #define VP8_MC_FUNC(IDX, SIZE, OPT) \
223 c->put_vp8_epel_pixels_tab[IDX][0][1] = ff_put_vp8_epel ## SIZE ## _h4_ ## OPT; \
224 c->put_vp8_epel_pixels_tab[IDX][1][0] = ff_put_vp8_epel ## SIZE ## _v4_ ## OPT; \
225 c->put_vp8_epel_pixels_tab[IDX][1][1] = ff_put_vp8_epel ## SIZE ## _h4v4_ ## OPT; \
226 c->put_vp8_epel_pixels_tab[IDX][1][2] = ff_put_vp8_epel ## SIZE ## _h6v4_ ## OPT; \
227 c->put_vp8_epel_pixels_tab[IDX][2][1] = ff_put_vp8_epel ## SIZE ## _h4v6_ ## OPT; \
228 VP8_LUMA_MC_FUNC(IDX, SIZE, OPT)
229
230 #define VP8_BILINEAR_MC_FUNC(IDX, SIZE, OPT) \
231 c->put_vp8_bilinear_pixels_tab[IDX][0][1] = ff_put_vp8_bilinear ## SIZE ## _h_ ## OPT; \
232 c->put_vp8_bilinear_pixels_tab[IDX][0][2] = ff_put_vp8_bilinear ## SIZE ## _h_ ## OPT; \
233 c->put_vp8_bilinear_pixels_tab[IDX][1][0] = ff_put_vp8_bilinear ## SIZE ## _v_ ## OPT; \
234 c->put_vp8_bilinear_pixels_tab[IDX][1][1] = ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT; \
235 c->put_vp8_bilinear_pixels_tab[IDX][1][2] = ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT; \
236 c->put_vp8_bilinear_pixels_tab[IDX][2][0] = ff_put_vp8_bilinear ## SIZE ## _v_ ## OPT; \
237 c->put_vp8_bilinear_pixels_tab[IDX][2][1] = ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT; \
238 c->put_vp8_bilinear_pixels_tab[IDX][2][2] = ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT
239
240
241 79 av_cold void ff_vp78dsp_init_x86(VP8DSPContext *c)
242 {
243 79 int cpu_flags = av_get_cpu_flags();
244
245
2/2
✓ Branch 0 taken 21 times.
✓ Branch 1 taken 58 times.
 79 if (EXTERNAL_SSE(cpu_flags)) {
246 21 c->put_vp8_epel_pixels_tab[0][0][0] =
247 21 c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_sse;
248 }
249
250
2/2
✓ Branch 0 taken 20 times.
✓ Branch 1 taken 59 times.
 79 if (EXTERNAL_SSE2_SLOW(cpu_flags)) {
251 20 c->put_vp8_epel_pixels_tab[1][0][0] =
252 20 c->put_vp8_bilinear_pixels_tab[1][0][0] = ff_put_vp8_pixels8_sse2;
253 20 VP8_LUMA_MC_FUNC(0, 16, sse2);
254 20 VP8_MC_FUNC(1, 8, sse2);
255 20 VP8_BILINEAR_MC_FUNC(0, 16, sse2);
256 20 VP8_BILINEAR_MC_FUNC(1, 8, sse2);
257 }
258
259 /* note that 4-tap width=16 functions are missing because w=16
260 * is only used for luma, and luma is always a copy or sixtap. */
261
2/2
✓ Branch 0 taken 18 times.
✓ Branch 1 taken 61 times.
 79 if (EXTERNAL_SSSE3(cpu_flags)) {
262 18 VP8_LUMA_MC_FUNC(0, 16, ssse3);
263 18 VP8_MC_FUNC(1, 8, ssse3);
264 18 VP8_MC_FUNC(2, 4, ssse3);
265 18 VP8_BILINEAR_MC_FUNC(0, 16, ssse3);
266 18 VP8_BILINEAR_MC_FUNC(1, 8, ssse3);
267 18 VP8_BILINEAR_MC_FUNC(2, 4, ssse3);
268 }
269 79 }
270
271 76 av_cold void ff_vp8dsp_init_x86(VP8DSPContext *c)
272 {
273 76 int cpu_flags = av_get_cpu_flags();
274
275
2/2
✓ Branch 0 taken 23 times.
✓ Branch 1 taken 53 times.
 76 if (EXTERNAL_MMX(cpu_flags)) {
276 23 c->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_mmx;
277 }
278
279
2/2
✓ Branch 0 taken 21 times.
✓ Branch 1 taken 55 times.
 76 if (EXTERNAL_SSE(cpu_flags)) {
280 21 c->vp8_idct_add = ff_vp8_idct_add_sse;
281 21 c->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_sse;
282 }
283
284
2/2
✓ Branch 0 taken 20 times.
✓ Branch 1 taken 56 times.
 76 if (EXTERNAL_SSE2_SLOW(cpu_flags)) {
285 20 c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_sse2;
286
287 20 c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_sse2;
288 20 c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_sse2;
289
290 20 c->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16y_mbedge_sse2;
291 20 c->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_mbedge_sse2;
292 }
293
294
2/2
✓ Branch 0 taken 20 times.
✓ Branch 1 taken 56 times.
 76 if (EXTERNAL_SSE2(cpu_flags)) {
295 20 c->vp8_idct_dc_add = ff_vp8_idct_dc_add_sse2;
296 20 c->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_sse2;
297
298 20 c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse2;
299
300 20 c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_sse2;
301 20 c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_sse2;
302
303 20 c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_sse2;
304 20 c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_sse2;
305 }
306
307
2/2
✓ Branch 0 taken 18 times.
✓ Branch 1 taken 58 times.
 76 if (EXTERNAL_SSSE3(cpu_flags)) {
308 18 c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_ssse3;
309 18 c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_ssse3;
310
311 18 c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_ssse3;
312 18 c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_ssse3;
313 18 c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_ssse3;
314 18 c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_ssse3;
315
316 18 c->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16y_mbedge_ssse3;
317 18 c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_ssse3;
318 18 c->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_mbedge_ssse3;
319 18 c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_ssse3;
320 }
321
322
2/2
✓ Branch 0 taken 17 times.
✓ Branch 1 taken 59 times.
 76 if (EXTERNAL_SSE4(cpu_flags)) {
323 17 c->vp8_idct_dc_add = ff_vp8_idct_dc_add_sse4;
324
325 17 c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse4;
326 17 c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_sse4;
327 17 c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_sse4;
328 }
329 76 }
330