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
178
179
180
181
|
/**
* Lightweight wrappers around AltiVec for 32-bit operations
* (C) 2009 Jack Lloyd
*
* Distributed under the terms of the Botan license
*/
#ifndef BOTAN_SIMD_ALTIVEC_H__
#define BOTAN_SIMD_ALTIVEC_H__
#include <botan/loadstor.h>
#include <altivec.h>
#undef vector
namespace Botan {
class SIMD_Altivec
{
public:
SIMD_Altivec(const u32bit B[4])
{
reg = (__vector unsigned int){B[0], B[1], B[2], B[3]};
}
SIMD_Altivec(u32bit B0, u32bit B1, u32bit B2, u32bit B3)
{
reg = (__vector unsigned int){B0, B1, B2, B3};
}
SIMD_Altivec(u32bit B)
{
reg = (__vector unsigned int){B, B, B, B};
}
static SIMD_Altivec load_le(const void* in)
{
const u32bit* in_32 = static_cast<const u32bit*>(in);
__vector unsigned int R0 = vec_ld(0, in_32);
__vector unsigned int R1 = vec_ld(12, in_32);
__vector unsigned char perm = vec_lvsl(0, in_32);
perm = vec_xor(perm, vec_splat_u8(3));
R0 = vec_perm(R0, R1, perm);
return SIMD_Altivec(R0);
}
static SIMD_Altivec load_be(const void* in)
{
const u32bit* in_32 = static_cast<const u32bit*>(in);
__vector unsigned int R0 = vec_ld(0, in_32);
__vector unsigned int R1 = vec_ld(12, in_32);
__vector unsigned char perm = vec_lvsl(0, in_32);
R0 = vec_perm(R0, R1, perm);
return SIMD_Altivec(R0);
}
void store_le(byte out[]) const
{
u32bit* out_32 = reinterpret_cast<u32bit*>(out);
__vector unsigned char perm = vec_lvsl(0, (int*)0);
perm = vec_xor(perm, vec_splat_u8(3));
__vector unsigned int swapped = vec_perm(reg, reg, perm);
vec_st(swapped, 0, out_32);
}
void store_be(byte out[]) const
{
u32bit* out_32 = reinterpret_cast<u32bit*>(out);
vec_st(reg, 0, out_32);
}
void rotate_left(u32bit rot)
{
__vector unsigned int rot_vec =
(__vector unsigned int){rot, rot, rot, rot};
reg = vec_rl(reg, rot_vec);
}
void rotate_right(u32bit rot)
{
rotate_left(32 - rot);
}
void operator+=(const SIMD_Altivec& other)
{
reg = vec_add(reg, other.reg);
}
SIMD_Altivec operator+(const SIMD_Altivec& other) const
{
return vec_add(reg, other.reg);
}
void operator-=(const SIMD_Altivec& other)
{
reg = vec_sub(reg, other.reg);
}
SIMD_Altivec operator-(const SIMD_Altivec& other) const
{
return vec_sub(reg, other.reg);
}
void operator^=(const SIMD_Altivec& other)
{
reg = vec_xor(reg, other.reg);
}
SIMD_Altivec operator^(const SIMD_Altivec& other) const
{
return vec_xor(reg, other.reg);
}
void operator|=(const SIMD_Altivec& other)
{
reg = vec_or(reg, other.reg);
}
void operator&=(const SIMD_Altivec& other)
{
reg = vec_and(reg, other.reg);
}
SIMD_Altivec operator<<(u32bit shift) const
{
__vector unsigned int shift_vec =
(__vector unsigned int){shift, shift, shift, shift};
return vec_sl(reg, shift_vec);
}
SIMD_Altivec operator>>(u32bit shift) const
{
__vector unsigned int shift_vec =
(__vector unsigned int){shift, shift, shift, shift};
return vec_sr(reg, shift_vec);
}
SIMD_Altivec operator~() const
{
return vec_nor(reg, reg);
}
static void transpose(SIMD_Altivec& B0, SIMD_Altivec& B1,
SIMD_Altivec& B2, SIMD_Altivec& B3)
{
__vector unsigned int T0 = vec_mergeh(B0.reg, B2.reg);
__vector unsigned int T1 = vec_mergel(B0.reg, B2.reg);
__vector unsigned int T2 = vec_mergeh(B1.reg, B3.reg);
__vector unsigned int T3 = vec_mergel(B1.reg, B3.reg);
B0.reg = vec_mergeh(T0, T2);
B1.reg = vec_mergel(T0, T2);
B2.reg = vec_mergeh(T1, T3);
B3.reg = vec_mergel(T1, T3);
}
private:
SIMD_Altivec(__vector unsigned int input) { reg = input; }
__vector unsigned int reg;
};
}
#endif
|
