1 /*
2 * Copyright (c) 2018 Paul B Mahol
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 "libavutil/attributes.h"
22 #include "libavutil/cpu.h"
23 #include "libavutil/x86/cpu.h"
24 #include "libavfilter/vf_overlay.h"
25
26 int ff_overlay_row_44_sse4(uint8_t *d, uint8_t *da, uint8_t *s, uint8_t *a,
27 int w, ptrdiff_t alinesize);
28
29 int ff_overlay_row_20_sse4(uint8_t *d, uint8_t *da, uint8_t *s, uint8_t *a,
30 int w, ptrdiff_t alinesize);
31
32 int ff_overlay_row_22_sse4(uint8_t *d, uint8_t *da, uint8_t *s, uint8_t *a,
33 int w, ptrdiff_t alinesize);
34
ff_overlay_init_x86(OverlayContext * s,int format,int pix_format,int alpha_format,int main_has_alpha)35 av_cold void ff_overlay_init_x86(OverlayContext *s, int format, int pix_format,
36 int alpha_format, int main_has_alpha)
37 {
38 int cpu_flags = av_get_cpu_flags();
39
40 if (EXTERNAL_SSE4(cpu_flags) &&
41 (format == OVERLAY_FORMAT_YUV444 ||
42 format == OVERLAY_FORMAT_GBRP) &&
43 alpha_format == 0 && main_has_alpha == 0) {
44 s->blend_row[0] = ff_overlay_row_44_sse4;
45 s->blend_row[1] = ff_overlay_row_44_sse4;
46 s->blend_row[2] = ff_overlay_row_44_sse4;
47 }
48
49 if (EXTERNAL_SSE4(cpu_flags) &&
50 (pix_format == AV_PIX_FMT_YUV420P) &&
51 (format == OVERLAY_FORMAT_YUV420) &&
52 alpha_format == 0 && main_has_alpha == 0) {
53 s->blend_row[0] = ff_overlay_row_44_sse4;
54 s->blend_row[1] = ff_overlay_row_20_sse4;
55 s->blend_row[2] = ff_overlay_row_20_sse4;
56 }
57
58 if (EXTERNAL_SSE4(cpu_flags) &&
59 (format == OVERLAY_FORMAT_YUV422) &&
60 alpha_format == 0 && main_has_alpha == 0) {
61 s->blend_row[0] = ff_overlay_row_44_sse4;
62 s->blend_row[1] = ff_overlay_row_22_sse4;
63 s->blend_row[2] = ff_overlay_row_22_sse4;
64 }
65 }
66