1 /*
2  * MMX/SSE/AVX constants used across x86 dsp optimizations.
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/mem_internal.h"
22 #include "libavutil/x86/asm.h" // for xmm_reg
23 #include "constants.h"
24 
25 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_1)    = { 0x0001000100010001ULL, 0x0001000100010001ULL,
26                                                     0x0001000100010001ULL, 0x0001000100010001ULL };
27 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_2)    = { 0x0002000200020002ULL, 0x0002000200020002ULL,
28                                                     0x0002000200020002ULL, 0x0002000200020002ULL };
29 DECLARE_ASM_ALIGNED(16, const xmm_reg,  ff_pw_3)    = { 0x0003000300030003ULL, 0x0003000300030003ULL };
30 DECLARE_ASM_ALIGNED(32, const ymm_reg,  ff_pw_4)    = { 0x0004000400040004ULL, 0x0004000400040004ULL,
31                                                     0x0004000400040004ULL, 0x0004000400040004ULL };
32 DECLARE_ASM_ALIGNED(16, const xmm_reg,  ff_pw_5)    = { 0x0005000500050005ULL, 0x0005000500050005ULL };
33 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_8)    = { 0x0008000800080008ULL, 0x0008000800080008ULL };
34 DECLARE_ASM_ALIGNED(16, const xmm_reg,  ff_pw_9)    = { 0x0009000900090009ULL, 0x0009000900090009ULL };
35 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_15)   =   0x000F000F000F000FULL;
36 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_16)   = { 0x0010001000100010ULL, 0x0010001000100010ULL };
37 DECLARE_ASM_ALIGNED(16, const xmm_reg,  ff_pw_18)   = { 0x0012001200120012ULL, 0x0012001200120012ULL };
38 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_20)   = { 0x0014001400140014ULL, 0x0014001400140014ULL };
39 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_32)   = { 0x0020002000200020ULL, 0x0020002000200020ULL };
40 DECLARE_ASM_ALIGNED(8,  const uint64_t, ff_pw_42)   =   0x002A002A002A002AULL;
41 DECLARE_ASM_ALIGNED(8,  const uint64_t, ff_pw_53)   =   0x0035003500350035ULL;
42 DECLARE_ASM_ALIGNED(16, const xmm_reg,  ff_pw_64)   = { 0x0040004000400040ULL, 0x0040004000400040ULL };
43 DECLARE_ASM_ALIGNED(8,  const uint64_t, ff_pw_96)   =   0x0060006000600060ULL;
44 DECLARE_ASM_ALIGNED(8,  const uint64_t, ff_pw_128)  =   0x0080008000800080ULL;
45 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_255)  = { 0x00ff00ff00ff00ffULL, 0x00ff00ff00ff00ffULL,
46                                                     0x00ff00ff00ff00ffULL, 0x00ff00ff00ff00ffULL };
47 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_256)  = { 0x0100010001000100ULL, 0x0100010001000100ULL,
48                                                     0x0100010001000100ULL, 0x0100010001000100ULL };
49 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_512)  = { 0x0200020002000200ULL, 0x0200020002000200ULL,
50                                                     0x0200020002000200ULL, 0x0200020002000200ULL };
51 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_1019) = { 0x03FB03FB03FB03FBULL, 0x03FB03FB03FB03FBULL };
52 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_1023) = { 0x03ff03ff03ff03ffULL, 0x03ff03ff03ff03ffULL,
53                                                     0x03ff03ff03ff03ffULL, 0x03ff03ff03ff03ffULL};
54 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_1024) = { 0x0400040004000400ULL, 0x0400040004000400ULL,
55                                                     0x0400040004000400ULL, 0x0400040004000400ULL};
56 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_2048) = { 0x0800080008000800ULL, 0x0800080008000800ULL,
57                                                     0x0800080008000800ULL, 0x0800080008000800ULL };
58 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_4095) = { 0x0fff0fff0fff0fffULL, 0x0fff0fff0fff0fffULL,
59                                                     0x0fff0fff0fff0fffULL, 0x0fff0fff0fff0fffULL };
60 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_4096) = { 0x1000100010001000ULL, 0x1000100010001000ULL,
61                                                     0x1000100010001000ULL, 0x1000100010001000ULL };
62 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_8192) = { 0x2000200020002000ULL, 0x2000200020002000ULL,
63                                                     0x2000200020002000ULL, 0x2000200020002000ULL };
64 DECLARE_ALIGNED(32, const ymm_reg,  ff_pw_m1)   = { 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL,
65                                                     0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL };
66 
67 DECLARE_ALIGNED(32, const ymm_reg,  ff_pb_0)    = { 0x0000000000000000ULL, 0x0000000000000000ULL,
68                                                     0x0000000000000000ULL, 0x0000000000000000ULL };
69 DECLARE_ALIGNED(32, const ymm_reg,  ff_pb_1)    = { 0x0101010101010101ULL, 0x0101010101010101ULL,
70                                                     0x0101010101010101ULL, 0x0101010101010101ULL };
71 DECLARE_ALIGNED(32, const ymm_reg,  ff_pb_2)    = { 0x0202020202020202ULL, 0x0202020202020202ULL,
72                                                     0x0202020202020202ULL, 0x0202020202020202ULL };
73 DECLARE_ALIGNED(32, const ymm_reg,  ff_pb_3)    = { 0x0303030303030303ULL, 0x0303030303030303ULL,
74                                                     0x0303030303030303ULL, 0x0303030303030303ULL };
75 DECLARE_ALIGNED(32, const xmm_reg,  ff_pb_15)   = { 0x0F0F0F0F0F0F0F0FULL, 0x0F0F0F0F0F0F0F0FULL };
76 DECLARE_ALIGNED(32, const ymm_reg,  ff_pb_80)   = { 0x8080808080808080ULL, 0x8080808080808080ULL,
77                                                     0x8080808080808080ULL, 0x8080808080808080ULL };
78 DECLARE_ALIGNED(32, const ymm_reg,  ff_pb_FE)   = { 0xFEFEFEFEFEFEFEFEULL, 0xFEFEFEFEFEFEFEFEULL,
79                                                     0xFEFEFEFEFEFEFEFEULL, 0xFEFEFEFEFEFEFEFEULL };
80 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_FC)   =   0xFCFCFCFCFCFCFCFCULL;
81 
82 DECLARE_ALIGNED(16, const xmm_reg,  ff_ps_neg)  = { 0x8000000080000000ULL, 0x8000000080000000ULL };
83 
84 DECLARE_ALIGNED(32, const ymm_reg,  ff_pd_1)    = { 0x0000000100000001ULL, 0x0000000100000001ULL,
85                                                     0x0000000100000001ULL, 0x0000000100000001ULL };
86 DECLARE_ALIGNED(32, const ymm_reg,  ff_pd_16)   = { 0x0000001000000010ULL, 0x0000001000000010ULL,
87                                                     0x0000001000000010ULL, 0x0000001000000010ULL };
88 DECLARE_ALIGNED(32, const ymm_reg,  ff_pd_32)   = { 0x0000002000000020ULL, 0x0000002000000020ULL,
89                                                     0x0000002000000020ULL, 0x0000002000000020ULL };
90 DECLARE_ALIGNED(32, const ymm_reg,  ff_pd_8192) = { 0x0000200000002000ULL, 0x0000200000002000ULL,
91                                                     0x0000200000002000ULL, 0x0000200000002000ULL };
92 DECLARE_ALIGNED(32, const ymm_reg,  ff_pd_65535)= { 0x0000ffff0000ffffULL, 0x0000ffff0000ffffULL,
93                                                     0x0000ffff0000ffffULL, 0x0000ffff0000ffffULL };
94