1 /*
2 * Lossless video DSP utils
3 *
4 * This file is part of FFmpeg.
5 *
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
10 *
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 */
20
21 #include "config.h"
22 #include "../lossless_videodsp.h"
23 #include "libavutil/x86/cpu.h"
24
25 void ff_add_bytes_sse2(uint8_t *dst, uint8_t *src, ptrdiff_t w);
26 void ff_add_bytes_avx2(uint8_t *dst, uint8_t *src, ptrdiff_t w);
27
28 void ff_add_median_pred_sse2(uint8_t *dst, const uint8_t *top,
29 const uint8_t *diff, ptrdiff_t w,
30 int *left, int *left_top);
31
32 int ff_add_left_pred_ssse3(uint8_t *dst, const uint8_t *src,
33 ptrdiff_t w, int left);
34 int ff_add_left_pred_unaligned_ssse3(uint8_t *dst, const uint8_t *src,
35 ptrdiff_t w, int left);
36 int ff_add_left_pred_unaligned_avx2(uint8_t *dst, const uint8_t *src,
37 ptrdiff_t w, int left);
38
39 int ff_add_left_pred_int16_ssse3(uint16_t *dst, const uint16_t *src, unsigned mask, ptrdiff_t w, unsigned acc);
40 int ff_add_left_pred_int16_unaligned_ssse3(uint16_t *dst, const uint16_t *src, unsigned mask, ptrdiff_t w, unsigned acc);
41
42 void ff_add_gradient_pred_ssse3(uint8_t *src, const ptrdiff_t stride, const ptrdiff_t width);
43 void ff_add_gradient_pred_avx2(uint8_t *src, const ptrdiff_t stride, const ptrdiff_t width);
44
ff_llviddsp_init_x86(LLVidDSPContext * c)45 void ff_llviddsp_init_x86(LLVidDSPContext *c)
46 {
47 int cpu_flags = av_get_cpu_flags();
48
49 if (EXTERNAL_SSE2(cpu_flags)) {
50 c->add_bytes = ff_add_bytes_sse2;
51 c->add_median_pred = ff_add_median_pred_sse2;
52 }
53
54 if (EXTERNAL_SSSE3(cpu_flags)) {
55 c->add_left_pred = ff_add_left_pred_ssse3;
56 c->add_left_pred_int16 = ff_add_left_pred_int16_ssse3;
57 c->add_gradient_pred = ff_add_gradient_pred_ssse3;
58 }
59
60 if (EXTERNAL_SSSE3_FAST(cpu_flags)) {
61 c->add_left_pred = ff_add_left_pred_unaligned_ssse3;
62 c->add_left_pred_int16 = ff_add_left_pred_int16_unaligned_ssse3;
63 }
64
65 if (EXTERNAL_AVX2_FAST(cpu_flags)) {
66 c->add_bytes = ff_add_bytes_avx2;
67 c->add_left_pred = ff_add_left_pred_unaligned_avx2;
68 c->add_gradient_pred = ff_add_gradient_pred_avx2;
69 }
70 }
71