aboutsummaryrefslogtreecommitdiffstats
path: root/libavutil/ppc/util_altivec.h
blob: 5527740010657f138268ec066e8556bb85f21fe4 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
/*
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

/**
 * @file
 * Contains misc utility macros and inline functions
 */

#ifndef AVUTIL_PPC_UTIL_ALTIVEC_H
#define AVUTIL_PPC_UTIL_ALTIVEC_H

#include <stdint.h>

#include "config.h"

#if HAVE_ALTIVEC_H
#include <altivec.h>
#endif

#include "types_altivec.h"

#if HAVE_ALTIVEC

// used to build registers permutation vectors (vcprm)
// the 's' are for words in the _s_econd vector
#define WORD_0 0x00,0x01,0x02,0x03
#define WORD_1 0x04,0x05,0x06,0x07
#define WORD_2 0x08,0x09,0x0a,0x0b
#define WORD_3 0x0c,0x0d,0x0e,0x0f
#define WORD_s0 0x10,0x11,0x12,0x13
#define WORD_s1 0x14,0x15,0x16,0x17
#define WORD_s2 0x18,0x19,0x1a,0x1b
#define WORD_s3 0x1c,0x1d,0x1e,0x1f
#define vcprm(a,b,c,d) (const vec_u8){WORD_ ## a, WORD_ ## b, WORD_ ## c, WORD_ ## d}

#define SWP_W2S0 0x02,0x03,0x00,0x01
#define SWP_W2S1 0x06,0x07,0x04,0x05
#define SWP_W2S2 0x0a,0x0b,0x08,0x09
#define SWP_W2S3 0x0e,0x0f,0x0c,0x0d
#define SWP_W2Ss0 0x12,0x13,0x10,0x11
#define SWP_W2Ss1 0x16,0x17,0x14,0x15
#define SWP_W2Ss2 0x1a,0x1b,0x18,0x19
#define SWP_W2Ss3 0x1e,0x1f,0x1c,0x1d
#define vcswapi2s(a,b,c,d) (const vector unsigned char){SWP_W2S ## a, SWP_W2S ## b, SWP_W2S ## c, SWP_W2S ## d}

#define vcswapc() \
  (const vector unsigned char){0x0f,0x0e,0x0d,0x0c,0x0b,0x0a,0x09,0x08,0x07,0x06,0x05,0x04,0x03,0x02,0x01,0x00}


// Transpose 8x8 matrix of 16-bit elements (in-place)
#define TRANSPOSE8(a,b,c,d,e,f,g,h) \
do { \
    vec_s16 A1, B1, C1, D1, E1, F1, G1, H1; \
    vec_s16 A2, B2, C2, D2, E2, F2, G2, H2; \
 \
    A1 = vec_mergeh (a, e); \
    B1 = vec_mergel (a, e); \
    C1 = vec_mergeh (b, f); \
    D1 = vec_mergel (b, f); \
    E1 = vec_mergeh (c, g); \
    F1 = vec_mergel (c, g); \
    G1 = vec_mergeh (d, h); \
    H1 = vec_mergel (d, h); \
 \
    A2 = vec_mergeh (A1, E1); \
    B2 = vec_mergel (A1, E1); \
    C2 = vec_mergeh (B1, F1); \
    D2 = vec_mergel (B1, F1); \
    E2 = vec_mergeh (C1, G1); \
    F2 = vec_mergel (C1, G1); \
    G2 = vec_mergeh (D1, H1); \
    H2 = vec_mergel (D1, H1); \
 \
    a = vec_mergeh (A2, E2); \
    b = vec_mergel (A2, E2); \
    c = vec_mergeh (B2, F2); \
    d = vec_mergel (B2, F2); \
    e = vec_mergeh (C2, G2); \
    f = vec_mergel (C2, G2); \
    g = vec_mergeh (D2, H2); \
    h = vec_mergel (D2, H2); \
} while (0)


#if HAVE_BIGENDIAN
#define VEC_LD(offset,b)                                   \
    vec_perm(vec_ld(offset, b), vec_ld((offset)+15, b), vec_lvsl(offset, b))
#else
#define VEC_LD(offset,b)                                   \
    vec_vsx_ld(offset, b)
#endif

/** @brief loads unaligned vector @a *src with offset @a offset
    and returns it */
#if HAVE_BIGENDIAN
static inline vec_u8 unaligned_load(int offset, const uint8_t *src)
{
    register vec_u8 first = vec_ld(offset, src);
    register vec_u8 second = vec_ld(offset + 15, src);
    register vec_u8 mask = vec_lvsl(offset, src);
    return vec_perm(first, second, mask);
}
static inline vec_u8 load_with_perm_vec(int offset, const uint8_t *src, vec_u8 perm_vec)
{
    vec_u8 a = vec_ld(offset, src);
    vec_u8 b = vec_ld(offset + 15, src);
    return vec_perm(a, b, perm_vec);
}
#else
#define unaligned_load(a,b) VEC_LD(a,b)
#define load_with_perm_vec(a,b,c) VEC_LD(a,b)
#endif


/**
 * loads vector known misalignment
 * @param perm_vec the align permute vector to combine the two loads from lvsl
 */

#define vec_unaligned_load(b)  VEC_LD(0, b)

#if HAVE_BIGENDIAN
#define VEC_MERGEH(a, b) vec_mergeh(a, b)
#define VEC_MERGEL(a, b) vec_mergel(a, b)
#else
#define VEC_MERGEH(a, b) vec_mergeh(b, a)
#define VEC_MERGEL(a, b) vec_mergel(b, a)
#endif

#if HAVE_BIGENDIAN
#define VEC_ST(a,b,c) vec_st(a,b,c)
#else
#define VEC_ST(a,b,c) vec_vsx_st(a,b,c)
#endif

#if HAVE_BIGENDIAN
#define VEC_SPLAT16(a,b) vec_splat((vec_s16)(a), b)
#else
#define VEC_SPLAT16(a,b) vec_splat((vec_s16)(vec_perm(a, a, vcswapi2s(0,1,2,3))), b)
#endif

#if HAVE_BIGENDIAN
#define VEC_SLD16(a,b,c) vec_sld(a, b, c)
#else
#define VEC_SLD16(a,b,c) vec_sld(b, a, c)
#endif

#endif /* HAVE_ALTIVEC */

#if HAVE_VSX
#if HAVE_BIGENDIAN
#define vsx_ld_u8_s16(off, p)                               \
    ((vec_s16)vec_mergeh((vec_u8)vec_splat_u8(0),           \
                         (vec_u8)vec_vsx_ld((off), (p))))
#else
#define vsx_ld_u8_s16(off, p)                               \
    ((vec_s16)vec_mergeh((vec_u8)vec_vsx_ld((off), (p)),    \
                         (vec_u8)vec_splat_u8(0)))
#endif /* HAVE_BIGENDIAN */
#endif /* HAVE_VSX */

#endif /* AVUTIL_PPC_UTIL_ALTIVEC_H */