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
|
/* This file is part of the gf2x library.
Copyright 2007, 2008, 2009, 2010, 2012, 2013, 2015
Richard Brent, Pierrick Gaudry, Emmanuel Thome', Paul Zimmermann
This program is free software; you can redistribute it and/or modify it
under the terms of either:
- If the archive contains a file named toom-gpl.c (not a trivial
placeholder), the GNU General Public License as published by the Free
Software Foundation; either version 3 of the License, or (at your
option) any later version.
- If the archive contains a file named toom-gpl.c which is a trivial
placeholder, 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.
This program 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 license text for more details.
You should have received a copy of the GNU General Public License as
well as the GNU Lesser General Public License along with this program;
see the files COPYING and COPYING.LIB. If not, write to the Free
Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
02110-1301, USA.
*/
/* Implements 128x128 -> 256 bit product using SSE2 instructions. */
#ifndef GF2X_MUL2_H_
#define GF2X_MUL2_H_
#include "gf2x.h"
/* All gf2x source files for lowlevel functions must include gf2x-small.h
* This is mandatory for the tuning mechanism. */
#include "gf2x/gf2x-small.h"
#if GF2X_WORDSIZE != 64
#error "This code is for 64-bit only"
#endif
#ifndef GF2X_HAVE_SSE2_SUPPORT
#error "This code needs sse-2 support"
#endif
#if defined(__GNUC__) && __GNUC__ == 4 && __GNUC_MINOR__ == 3 && (__GNUC_PATCHLEVEL__ == 0 || __GNUC_PATCHLEVEL__ == 1)
#warning "Your GCC version is buggy. Binary fields may fail randomly"
/* Gcc bug reports 37101 and 37340 -- the only convenient fix is to
* upgrade to 4.3.2 */
#endif
/* This code has been modified in comparison to the previously existing
* one, in order to use Intel intrinsics exclusively, and not rely on gcc
* syntax.
*
* The Intel C++ Intrinsics Reference is a good read. However I can't
* find the official link right now. An old version can be found in the wild:
*
* http://www.info.univ-angers.fr/~richer/ens/l3info/ao/intel_intrinsics.pdf
*
* I don't even know if this document evolved. Probably not.
*
* It says: ``Document Number: 312482-003US''
*/
GF2X_STORAGE_CLASS_mul2
void gf2x_mul2(unsigned long * t, unsigned long const * s1,
unsigned long const * s2)
{
#define SHL(x, r) _mm_slli_epi64((x), (r))
#define SHR(x, r) _mm_srli_epi64((x), (r))
#define SHLD(x, r) _mm_slli_si128((x), (r) >> 3)
#define SHRD(x, r) _mm_srli_si128((x), (r) >> 3)
#define XOREQ(lop, rop) lop = _mm_xor_si128((lop), (rop))
#define PXOR(lop, rop) _mm_xor_si128((lop), (rop))
#define PAND(lop, rop) _mm_and_si128((lop), (rop))
#define PZERO _mm_setzero_si128()
#define PSUB(x, y) _mm_sub_epi64((x), (y))
#define PNEG(x) PSUB(PZERO, (x))
__m128i u;
__m128i t0;
__m128i t1;
__m128i t2;
__m128i g[16];
__m128i w;
// __m128i m = _mm_set1_epi32(0xeeeeeeee);
__m128i m = _gf2x_mm_set1_epi64_c(0xeeeeeeeeeeeeeeee);
/* sequence update walk */
__m128i b0 = _mm_loadu_si128((__m128i*) s2);
g[ 0] = _mm_setzero_si128();
g[ 1] = b0;
__m128i v1 = _mm_loadu_si128((__m128i*) s1);
w = PNEG(SHR(b0,63));
__m128i v2 = _mm_unpackhi_epi64(v1, v1);
v1 = _mm_unpacklo_epi64(v1, v1);
v1 = SHR(PAND(v1, m), 1); t1 = PAND(v1, w);
g[ 2] = SHL(b0, 1); g[ 3] = PXOR(g[ 2], b0);
v2 = SHR(PAND(v2, m), 1); t2 = PAND(v2, w);
g[ 4] = SHL(g[ 2], 1); g[ 5] = PXOR(g[ 4], b0);
w = PNEG(SHR(g[ 2],63));
g[ 6] = SHL(g[ 3], 1); g[ 7] = PXOR(g[ 6], b0);
v1 = SHR(PAND(v1, m), 1); XOREQ(t1, PAND(v1, w));
g[ 8] = SHL(g[ 4], 1); g[ 9] = PXOR(g[ 8], b0);
v2 = SHR(PAND(v2, m), 1); XOREQ(t2, PAND(v2, w));
g[10] = SHL(g[ 5], 1); g[11] = PXOR(g[10], b0);
w = PNEG(SHR(g[4],63));
g[12] = SHL(g[ 6], 1); g[13] = PXOR(g[12], b0);
v1 = SHR(PAND(v1, m), 1); XOREQ(t1, PAND(v1, w));
g[14] = SHL(g[ 7], 1); g[15] = PXOR(g[14], b0);
v2 = SHR(PAND(v2, m), 1); XOREQ(t2, PAND(v2, w));
/* round 0 */
u = g[s1[0] & 15]; t0 = u;
u = g[s1[0] >> 4 & 15]; XOREQ(t0, SHL(u, 4)); XOREQ(t1, SHR(u, 60));
u = g[s1[0] >> 8 & 15]; XOREQ(t0, SHL(u, 8)); XOREQ(t1, SHR(u, 56));
u = g[s1[0] >> 12 & 15]; XOREQ(t0, SHL(u, 12)); XOREQ(t1, SHR(u, 52));
u = g[s1[0] >> 16 & 15]; XOREQ(t0, SHL(u, 16)); XOREQ(t1, SHR(u, 48));
u = g[s1[0] >> 20 & 15]; XOREQ(t0, SHL(u, 20)); XOREQ(t1, SHR(u, 44));
u = g[s1[0] >> 24 & 15]; XOREQ(t0, SHL(u, 24)); XOREQ(t1, SHR(u, 40));
u = g[s1[0] >> 28 & 15]; XOREQ(t0, SHL(u, 28)); XOREQ(t1, SHR(u, 36));
u = g[s1[0] >> 32 & 15]; XOREQ(t0, SHL(u, 32)); XOREQ(t1, SHR(u, 32));
u = g[s1[0] >> 36 & 15]; XOREQ(t0, SHL(u, 36)); XOREQ(t1, SHR(u, 28));
u = g[s1[0] >> 40 & 15]; XOREQ(t0, SHL(u, 40)); XOREQ(t1, SHR(u, 24));
u = g[s1[0] >> 44 & 15]; XOREQ(t0, SHL(u, 44)); XOREQ(t1, SHR(u, 20));
u = g[s1[0] >> 48 & 15]; XOREQ(t0, SHL(u, 48)); XOREQ(t1, SHR(u, 16));
u = g[s1[0] >> 52 & 15]; XOREQ(t0, SHL(u, 52)); XOREQ(t1, SHR(u, 12));
u = g[s1[0] >> 56 & 15]; XOREQ(t0, SHL(u, 56)); XOREQ(t1, SHR(u, 8));
u = g[s1[0] >> 60 & 15]; XOREQ(t0, SHL(u, 60)); XOREQ(t1, SHR(u, 4));
/* round 1 */
u = g[s1[1] & 15]; XOREQ(t1, u);
u = g[s1[1] >> 4 & 15]; XOREQ(t1, SHL(u, 4)); XOREQ(t2, SHR(u, 60));
u = g[s1[1] >> 8 & 15]; XOREQ(t1, SHL(u, 8)); XOREQ(t2, SHR(u, 56));
u = g[s1[1] >> 12 & 15]; XOREQ(t1, SHL(u, 12)); XOREQ(t2, SHR(u, 52));
u = g[s1[1] >> 16 & 15]; XOREQ(t1, SHL(u, 16)); XOREQ(t2, SHR(u, 48));
u = g[s1[1] >> 20 & 15]; XOREQ(t1, SHL(u, 20)); XOREQ(t2, SHR(u, 44));
u = g[s1[1] >> 24 & 15]; XOREQ(t1, SHL(u, 24)); XOREQ(t2, SHR(u, 40));
u = g[s1[1] >> 28 & 15]; XOREQ(t1, SHL(u, 28)); XOREQ(t2, SHR(u, 36));
u = g[s1[1] >> 32 & 15]; XOREQ(t1, SHL(u, 32)); XOREQ(t2, SHR(u, 32));
u = g[s1[1] >> 36 & 15]; XOREQ(t1, SHL(u, 36)); XOREQ(t2, SHR(u, 28));
u = g[s1[1] >> 40 & 15]; XOREQ(t1, SHL(u, 40)); XOREQ(t2, SHR(u, 24));
u = g[s1[1] >> 44 & 15]; XOREQ(t1, SHL(u, 44)); XOREQ(t2, SHR(u, 20));
u = g[s1[1] >> 48 & 15]; XOREQ(t1, SHL(u, 48)); XOREQ(t2, SHR(u, 16));
u = g[s1[1] >> 52 & 15]; XOREQ(t1, SHL(u, 52)); XOREQ(t2, SHR(u, 12));
u = g[s1[1] >> 56 & 15]; XOREQ(t1, SHL(u, 56)); XOREQ(t2, SHR(u, 8));
u = g[s1[1] >> 60 & 15]; XOREQ(t1, SHL(u, 60)); XOREQ(t2, SHR(u, 4));
/* end */
/* store result */
_mm_storeu_si128((__m128i*)t, PXOR(t0, SHLD(t1, 64)));
_mm_storeu_si128((__m128i*)(t+2), PXOR(t2, SHRD(t1, 64)));
#undef PNEG
#undef PSUB
#undef PZERO
#undef SHL
#undef SHR
#undef SHLD
#undef SHRD
#undef XOREQ
#undef PXOR
#undef PAND
}
#endif /* GF2X_MUL2_H_ */
|
