-
Notifications
You must be signed in to change notification settings - Fork 13
/
KeccakP-1600-AVX512-plainC.c
253 lines (219 loc) · 9.25 KB
/
KeccakP-1600-AVX512-plainC.c
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
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
/*
K12 based on the eXtended Keccak Code Package (XKCP)
https://github.com/XKCP/XKCP
The Keccak-p permutations, designed by Guido Bertoni, Joan Daemen, Michaël Peeters and Gilles Van Assche.
Implementation by Ronny Van Keer, hereby denoted as "the implementer".
For more information, feedback or questions, please refer to the Keccak Team website:
https://keccak.team/
To the extent possible under law, the implementer has waived all copyright
and related or neighboring rights to the source code in this file.
http://creativecommons.org/publicdomain/zero/1.0/
---
We would like to thank Vladimir Sedach, we have used parts of his Keccak AVX-512 C++ code.
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdint.h>
#include <smmintrin.h>
#include <wmmintrin.h>
#include <immintrin.h>
#include <emmintrin.h>
#include "align.h"
typedef __m512i V512;
#define XOR(a,b) _mm512_xor_si512(a,b)
#define XOR3(a,b,c) _mm512_ternarylogic_epi64(a,b,c,0x96)
#define XOR5(a,b,c,d,e) XOR3(XOR3(a,b,c),d,e)
#define ROL(a,offset) _mm512_rol_epi64(a,offset)
#define Chi(a,b,c) _mm512_ternarylogic_epi64(a,b,c,0xD2)
#define LOAD_Lanes(m,a) _mm512_maskz_loadu_epi64(m,a)
#define LOAD_Lane(a) LOAD_Lanes(0x01,a)
#define LOAD_Plane(a) LOAD_Lanes(0x1F,a)
#define LOAD_8Lanes(a) LOAD_Lanes(0xFF,a)
#define STORE_Lanes(a,m,v) _mm512_mask_storeu_epi64(a,m,v)
#define STORE_Lane(a,v) STORE_Lanes(a,0x01,v)
#define STORE_Plane(a,v) STORE_Lanes(a,0x1F,v)
#define STORE_8Lanes(a,v) STORE_Lanes(a,0xFF,v)
/* ---------------------------------------------------------------- */
void KeccakP1600_AVX512_Initialize(void *state)
{
memset(state, 0, 1600/8);
}
/* ---------------------------------------------------------------- */
void KeccakP1600_AVX512_AddBytes(void *state, const unsigned char *data, unsigned int offset, unsigned int length)
{
uint8_t *stateAsBytes;
uint64_t *stateAsLanes;
for( stateAsBytes = (uint8_t*)state; ((offset % 8) != 0) && (length != 0); ++offset, --length)
stateAsBytes[offset] ^= *(data++);
for (stateAsLanes = (uint64_t*)(stateAsBytes + offset); length >= 8*8; stateAsLanes += 8, data += 8*8, length -= 8*8)
STORE_8Lanes( stateAsLanes, XOR(LOAD_8Lanes(stateAsLanes), LOAD_8Lanes((const uint64_t*)data)));
for (/* empty */; length >= 8; ++stateAsLanes, data += 8, length -= 8)
STORE_Lane( stateAsLanes, XOR(LOAD_Lane(stateAsLanes), LOAD_Lane((const uint64_t*)data)));
for ( stateAsBytes = (uint8_t*)stateAsLanes; length != 0; --length)
*(stateAsBytes++) ^= *(data++);
}
/* ---------------------------------------------------------------- */
void KeccakP1600_AVX512_ExtractBytes(const void *state, unsigned char *data, unsigned int offset, unsigned int length)
{
memcpy(data, (unsigned char*)state+offset, length);
}
/* ---------------------------------------------------------------- */
const uint64_t KeccakP1600RoundConstants[24] = {
0x0000000000000001ULL,
0x0000000000008082ULL,
0x800000000000808aULL,
0x8000000080008000ULL,
0x000000000000808bULL,
0x0000000080000001ULL,
0x8000000080008081ULL,
0x8000000000008009ULL,
0x000000000000008aULL,
0x0000000000000088ULL,
0x0000000080008009ULL,
0x000000008000000aULL,
0x000000008000808bULL,
0x800000000000008bULL,
0x8000000000008089ULL,
0x8000000000008003ULL,
0x8000000000008002ULL,
0x8000000000000080ULL,
0x000000000000800aULL,
0x800000008000000aULL,
0x8000000080008081ULL,
0x8000000000008080ULL,
0x0000000080000001ULL,
0x8000000080008008ULL };
#define KeccakP_DeclareVars \
V512 b0, b1, b2, b3, b4; \
V512 Baeiou, Gaeiou, Kaeiou, Maeiou, Saeiou; \
V512 moveThetaPrev = _mm512_setr_epi64(4, 0, 1, 2, 3, 5, 6, 7); \
V512 moveThetaNext = _mm512_setr_epi64(1, 2, 3, 4, 0, 5, 6, 7); \
V512 rhoB = _mm512_setr_epi64( 0, 1, 62, 28, 27, 0, 0, 0); \
V512 rhoG = _mm512_setr_epi64(36, 44, 6, 55, 20, 0, 0, 0); \
V512 rhoK = _mm512_setr_epi64( 3, 10, 43, 25, 39, 0, 0, 0); \
V512 rhoM = _mm512_setr_epi64(41, 45, 15, 21, 8, 0, 0, 0); \
V512 rhoS = _mm512_setr_epi64(18, 2, 61, 56, 14, 0, 0, 0); \
V512 pi1B = _mm512_setr_epi64(0, 3, 1, 4, 2, 5, 6, 7); \
V512 pi1G = _mm512_setr_epi64(1, 4, 2, 0, 3, 5, 6, 7); \
V512 pi1K = _mm512_setr_epi64(2, 0, 3, 1, 4, 5, 6, 7); \
V512 pi1M = _mm512_setr_epi64(3, 1, 4, 2, 0, 5, 6, 7); \
V512 pi1S = _mm512_setr_epi64(4, 2, 0, 3, 1, 5, 6, 7); \
V512 pi2S1 = _mm512_setr_epi64(0, 1, 2, 3, 4, 5, 0+8, 2+8); \
V512 pi2S2 = _mm512_setr_epi64(0, 1, 2, 3, 4, 5, 1+8, 3+8); \
V512 pi2BG = _mm512_setr_epi64(0, 1, 0+8, 1+8, 6, 5, 6, 7); \
V512 pi2KM = _mm512_setr_epi64(2, 3, 2+8, 3+8, 7, 5, 6, 7); \
V512 pi2S3 = _mm512_setr_epi64(4, 5, 4+8, 5+8, 4, 5, 6, 7);
#define copyFromState(pState) \
Baeiou = LOAD_Plane(pState+ 0); \
Gaeiou = LOAD_Plane(pState+ 5); \
Kaeiou = LOAD_Plane(pState+10); \
Maeiou = LOAD_Plane(pState+15); \
Saeiou = LOAD_Plane(pState+20);
#define copyToState(pState) \
STORE_Plane(pState+ 0, Baeiou); \
STORE_Plane(pState+ 5, Gaeiou); \
STORE_Plane(pState+10, Kaeiou); \
STORE_Plane(pState+15, Maeiou); \
STORE_Plane(pState+20, Saeiou);
#define KeccakP_Round(i) \
/* Theta */ \
b0 = XOR5( Baeiou, Gaeiou, Kaeiou, Maeiou, Saeiou ); \
b1 = _mm512_permutexvar_epi64(moveThetaPrev, b0); \
b0 = _mm512_permutexvar_epi64(moveThetaNext, b0); \
b0 = _mm512_rol_epi64(b0, 1); \
Baeiou = XOR3( Baeiou, b0, b1 ); \
Gaeiou = XOR3( Gaeiou, b0, b1 ); \
Kaeiou = XOR3( Kaeiou, b0, b1 ); \
Maeiou = XOR3( Maeiou, b0, b1 ); \
Saeiou = XOR3( Saeiou, b0, b1 ); \
/* Rho */ \
Baeiou = _mm512_rolv_epi64(Baeiou, rhoB); \
Gaeiou = _mm512_rolv_epi64(Gaeiou, rhoG); \
Kaeiou = _mm512_rolv_epi64(Kaeiou, rhoK); \
Maeiou = _mm512_rolv_epi64(Maeiou, rhoM); \
Saeiou = _mm512_rolv_epi64(Saeiou, rhoS); \
/* Pi 1 */ \
b0 = _mm512_permutexvar_epi64(pi1B, Baeiou); \
b1 = _mm512_permutexvar_epi64(pi1G, Gaeiou); \
b2 = _mm512_permutexvar_epi64(pi1K, Kaeiou); \
b3 = _mm512_permutexvar_epi64(pi1M, Maeiou); \
b4 = _mm512_permutexvar_epi64(pi1S, Saeiou); \
/* Chi */ \
Baeiou = Chi(b0, b1, b2); \
Gaeiou = Chi(b1, b2, b3); \
Kaeiou = Chi(b2, b3, b4); \
Maeiou = Chi(b3, b4, b0); \
Saeiou = Chi(b4, b0, b1); \
/* Iota */ \
Baeiou = XOR(Baeiou, LOAD_Lane(KeccakP1600RoundConstants+i)); \
/* Pi 2 */ \
b0 = _mm512_unpacklo_epi64(Baeiou, Gaeiou); \
b1 = _mm512_unpacklo_epi64(Kaeiou, Maeiou); \
b0 = _mm512_permutex2var_epi64(b0, pi2S1, Saeiou); \
b2 = _mm512_unpackhi_epi64(Baeiou, Gaeiou); \
b3 = _mm512_unpackhi_epi64(Kaeiou, Maeiou); \
b2 = _mm512_permutex2var_epi64(b2, pi2S2, Saeiou); \
Baeiou = _mm512_permutex2var_epi64(b0, pi2BG, b1); \
Gaeiou = _mm512_permutex2var_epi64(b2, pi2BG, b3); \
Kaeiou = _mm512_permutex2var_epi64(b0, pi2KM, b1); \
Maeiou = _mm512_permutex2var_epi64(b2, pi2KM, b3); \
b0 = _mm512_permutex2var_epi64(b0, pi2S3, b1); \
Saeiou = _mm512_mask_blend_epi64(0x10, b0, Saeiou)
#define rounds12 \
KeccakP_Round( 12 ); \
KeccakP_Round( 13 ); \
KeccakP_Round( 14 ); \
KeccakP_Round( 15 ); \
KeccakP_Round( 16 ); \
KeccakP_Round( 17 ); \
KeccakP_Round( 18 ); \
KeccakP_Round( 19 ); \
KeccakP_Round( 20 ); \
KeccakP_Round( 21 ); \
KeccakP_Round( 22 ); \
KeccakP_Round( 23 )
/* ---------------------------------------------------------------- */
void KeccakP1600_AVX512_Permute_12rounds(void *state)
{
KeccakP_DeclareVars
uint64_t *stateAsLanes = (uint64_t*)state;
copyFromState(stateAsLanes);
rounds12;
copyToState(stateAsLanes);
}
/* ---------------------------------------------------------------- */
#include <assert.h>
size_t KeccakP1600_AVX512_12rounds_FastLoop_Absorb(void *state, unsigned int laneCount, const unsigned char *data, size_t dataByteLen)
{
size_t originalDataByteLen = dataByteLen;
assert(laneCount == 21 || laneCount == 17);
KeccakP_DeclareVars;
uint64_t *stateAsLanes = (uint64_t*)state;
uint64_t *inDataAsLanes = (uint64_t*)data;
if (laneCount == 21) {
#define laneCount 21
copyFromState(stateAsLanes);
while(dataByteLen >= 21*8) {
Baeiou = XOR(Baeiou, LOAD_Plane(inDataAsLanes+ 0));
Gaeiou = XOR(Gaeiou, LOAD_Plane(inDataAsLanes+ 5));
Kaeiou = XOR(Kaeiou, LOAD_Plane(inDataAsLanes+10));
Maeiou = XOR(Maeiou, LOAD_Plane(inDataAsLanes+15));
Saeiou = XOR(Saeiou, LOAD_Lane(inDataAsLanes+20));
rounds12;
inDataAsLanes += 21;
dataByteLen -= 21*8;
}
#undef laneCount
copyToState(stateAsLanes);
} else if (laneCount == 17) {
// TODO: further optimization needed for this case, laneCount == 17.
while(dataByteLen >= laneCount*8) {
KeccakP1600_AddBytes(state, data, 0, laneCount*8);
KeccakP1600_Permute_12rounds(state);
data += laneCount*8;
dataByteLen -= laneCount*8;
}
}
return originalDataByteLen - dataByteLen;
}