1 /*
2  * This file is part of FFmpeg.
3  *
4  * FFmpeg is free software; you can redistribute it and/or
5  * modify it under the terms of the GNU Lesser General Public
6  * License as published by the Free Software Foundation; either
7  * version 2.1 of the License, or (at your option) any later version.
8  *
9  * FFmpeg is distributed in the hope that it will be useful,
10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
12  * Lesser General Public License for more details.
13  *
14  * You should have received a copy of the GNU Lesser General Public
15  * License along with FFmpeg; if not, write to the Free Software
16  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17  */
18 
19 #include "config.h"
20 
21 #include "fdct.h"
22 #include "xvididct.h"
23 #include "simple_idct.h"
24 
25 #if ARCH_X86_64 && HAVE_MMX && HAVE_YASM
26 void ff_prores_idct_put_10_sse2(uint16_t *dst, int linesize,
27                                 int16_t *block, int16_t *qmat);
28 
ff_prores_idct_put_10_sse2_wrap(int16_t * dst)29 static void ff_prores_idct_put_10_sse2_wrap(int16_t *dst){
30     DECLARE_ALIGNED(16, static int16_t, qmat)[64];
31     DECLARE_ALIGNED(16, static int16_t, tmp)[64];
32     int i;
33 
34     for(i=0; i<64; i++){
35         qmat[i]=4;
36         tmp[i]= dst[i];
37     }
38     ff_prores_idct_put_10_sse2(dst, 16, tmp, qmat);
39 
40     for(i=0; i<64; i++) {
41          dst[i] -= 512;
42     }
43 }
44 #endif
45 
46 static const struct algo fdct_tab_arch[] = {
47 #if HAVE_MMX_INLINE
48     { "MMX",    ff_fdct_mmx,    FF_IDCT_PERM_NONE, AV_CPU_FLAG_MMX },
49 #endif
50 #if HAVE_MMXEXT_INLINE
51     { "MMXEXT", ff_fdct_mmxext, FF_IDCT_PERM_NONE, AV_CPU_FLAG_MMXEXT },
52 #endif
53 #if HAVE_SSE2_INLINE
54     { "SSE2",   ff_fdct_sse2,   FF_IDCT_PERM_NONE, AV_CPU_FLAG_SSE2 },
55 #endif
56     { 0 }
57 };
58 
59 static const struct algo idct_tab_arch[] = {
60 #if HAVE_MMX_INLINE
61     { "SIMPLE-MMX",  ff_simple_idct_mmx,  FF_IDCT_PERM_SIMPLE, AV_CPU_FLAG_MMX },
62 #endif
63 #if CONFIG_MPEG4_DECODER
64 #if HAVE_MMX_INLINE
65     { "XVID-MMX",    ff_xvid_idct_mmx,    FF_IDCT_PERM_NONE,   AV_CPU_FLAG_MMX,    1 },
66 #endif
67 #if HAVE_MMXEXT_INLINE
68     { "XVID-MMXEXT", ff_xvid_idct_mmxext, FF_IDCT_PERM_NONE,   AV_CPU_FLAG_MMXEXT, 1 },
69 #endif
70 #if HAVE_SSE2_INLINE
71     { "XVID-SSE2",   ff_xvid_idct_sse2,   FF_IDCT_PERM_SSE2,   AV_CPU_FLAG_SSE2,   1 },
72 #if ARCH_X86_64 && HAVE_YASM
73     { "PR-SSE2",     ff_prores_idct_put_10_sse2_wrap, FF_IDCT_PERM_TRANSPOSE, AV_CPU_FLAG_SSE2, 1 },
74 #endif
75 #endif
76 #endif /* CONFIG_MPEG4_DECODER */
77     { 0 }
78 };
79 
80 static short idct_simple_mmx_perm[64] = {
81     0x00, 0x08, 0x04, 0x09, 0x01, 0x0C, 0x05, 0x0D,
82     0x10, 0x18, 0x14, 0x19, 0x11, 0x1C, 0x15, 0x1D,
83     0x20, 0x28, 0x24, 0x29, 0x21, 0x2C, 0x25, 0x2D,
84     0x12, 0x1A, 0x16, 0x1B, 0x13, 0x1E, 0x17, 0x1F,
85     0x02, 0x0A, 0x06, 0x0B, 0x03, 0x0E, 0x07, 0x0F,
86     0x30, 0x38, 0x34, 0x39, 0x31, 0x3C, 0x35, 0x3D,
87     0x22, 0x2A, 0x26, 0x2B, 0x23, 0x2E, 0x27, 0x2F,
88     0x32, 0x3A, 0x36, 0x3B, 0x33, 0x3E, 0x37, 0x3F,
89 };
90 
91 static const uint8_t idct_sse2_row_perm[8] = { 0, 4, 1, 5, 2, 6, 3, 7 };
92 
permute_x86(int16_t dst[64],const int16_t src[64],enum idct_permutation_type perm_type)93 static int permute_x86(int16_t dst[64], const int16_t src[64],
94                        enum idct_permutation_type perm_type)
95 {
96     int i;
97 
98     switch (perm_type) {
99     case FF_IDCT_PERM_SIMPLE:
100         for (i = 0; i < 64; i++)
101             dst[idct_simple_mmx_perm[i]] = src[i];
102         return 1;
103     case FF_IDCT_PERM_SSE2:
104         for (i = 0; i < 64; i++)
105             dst[(i & 0x38) | idct_sse2_row_perm[i & 7]] = src[i];
106         return 1;
107     }
108 
109     return 0;
110 }
111