Go to most recent revision | Details | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
4349 | Serge | 1 | /* |
2 | * Copyright (C) 2009 Loren Merritt |
||
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 "libavutil/attributes.h" |
||
23 | #include "libavutil/cpu.h" |
||
24 | #include "libavutil/mem.h" |
||
25 | #include "libavutil/x86/asm.h" |
||
26 | #include "libavutil/x86/cpu.h" |
||
27 | #include "libavfilter/gradfun.h" |
||
28 | |||
29 | #if HAVE_YASM |
||
30 | void ff_gradfun_filter_line_mmxext(intptr_t x, uint8_t *dst, const uint8_t *src, |
||
31 | const uint16_t *dc, int thresh, |
||
32 | const uint16_t *dithers); |
||
33 | static void gradfun_filter_line_mmxext(uint8_t *dst, const uint8_t *src, |
||
34 | const uint16_t *dc, |
||
35 | int width, int thresh, |
||
36 | const uint16_t *dithers) |
||
37 | { |
||
38 | intptr_t x; |
||
39 | if (width & 3) { |
||
40 | x = width & ~3; |
||
41 | ff_gradfun_filter_line_c(dst + x, src + x, dc + x / 2, width - x, thresh, dithers); |
||
42 | width = x; |
||
43 | } |
||
44 | x = -width; |
||
45 | ff_gradfun_filter_line_mmxext(x, dst + width, src + width, dc + width/2, |
||
46 | thresh, dithers); |
||
47 | } |
||
48 | |||
49 | void ff_gradfun_filter_line_ssse3(intptr_t x, uint8_t *dst, const uint8_t *src, |
||
50 | const uint16_t *dc, int thresh, |
||
51 | const uint16_t *dithers); |
||
52 | static void gradfun_filter_line_ssse3(uint8_t *dst, const uint8_t *src, const uint16_t *dc, int width, int thresh, const uint16_t *dithers) |
||
53 | { |
||
54 | intptr_t x; |
||
55 | if (width & 7) { |
||
56 | // could be 10% faster if I somehow eliminated this |
||
57 | x = width & ~7; |
||
58 | ff_gradfun_filter_line_c(dst + x, src + x, dc + x / 2, width - x, thresh, dithers); |
||
59 | width = x; |
||
60 | } |
||
61 | x = -width; |
||
62 | ff_gradfun_filter_line_ssse3(x, dst + width, src + width, dc + width/2, |
||
63 | thresh, dithers); |
||
64 | } |
||
65 | |||
66 | void ff_gradfun_blur_line_movdqa_sse2(intptr_t x, uint16_t *buf, const uint16_t *buf1, uint16_t *dc, const uint8_t *src1, const uint8_t *src2); |
||
67 | void ff_gradfun_blur_line_movdqu_sse2(intptr_t x, uint16_t *buf, const uint16_t *buf1, uint16_t *dc, const uint8_t *src1, const uint8_t *src2); |
||
68 | static void gradfun_blur_line_sse2(uint16_t *dc, uint16_t *buf, const uint16_t *buf1, const uint8_t *src, int src_linesize, int width) |
||
69 | { |
||
70 | intptr_t x = -2*width; |
||
71 | if (((intptr_t) src | src_linesize) & 15) { |
||
72 | ff_gradfun_blur_line_movdqu_sse2(x, buf + width, buf1 + width, |
||
73 | dc + width, src + width * 2, |
||
74 | src + width * 2 + src_linesize); |
||
75 | } else { |
||
76 | ff_gradfun_blur_line_movdqa_sse2(x, buf + width, buf1 + width, |
||
77 | dc + width, src + width * 2, |
||
78 | src + width * 2 + src_linesize); |
||
79 | } |
||
80 | } |
||
81 | #endif /* HAVE_YASM */ |
||
82 | |||
83 | av_cold void ff_gradfun_init_x86(GradFunContext *gf) |
||
84 | { |
||
85 | #if HAVE_YASM |
||
86 | int cpu_flags = av_get_cpu_flags(); |
||
87 | |||
88 | if (EXTERNAL_MMXEXT(cpu_flags)) |
||
89 | gf->filter_line = gradfun_filter_line_mmxext; |
||
90 | if (EXTERNAL_SSSE3(cpu_flags)) |
||
91 | gf->filter_line = gradfun_filter_line_ssse3; |
||
92 | |||
93 | if (EXTERNAL_SSE2(cpu_flags)) |
||
94 | gf->blur_line = gradfun_blur_line_sse2; |
||
95 | #endif /* HAVE_YASM */ |
||
96 | } |