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
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
|
#!/usr/bin/env perl
# Copyright 2017-2023 The OpenSSL Project Authors. All Rights Reserved.
#
# Licensed under the Apache License 2.0 (the "License"). You may not use
# this file except in compliance with the License. You can obtain a copy
# in the file LICENSE in the source distribution or at
# https://www.openssl.org/source/license.html
#
# ====================================================================
# Written by Andy Polyakov, @dot-asm, initially for use in the OpenSSL
# project. The module is, however, dual licensed under OpenSSL and
# CRYPTOGAMS licenses depending on where you obtain it. For further
# details see https://github.com/dot-asm/cryptogams/.
# ====================================================================
#
# Keccak-1600 for AVX512VL.
#
# December 2017.
#
# This is an adaptation of AVX2 module that reuses register data
# layout, but utilizes new 256-bit AVX512VL instructions. See AVX2
# module for further information on layout.
#
########################################################################
# Numbers are cycles per processed byte out of large message.
#
# r=1088(*)
#
# Skylake-X 6.4/+47%
#
# (*) Corresponds to SHA3-256. Percentage after slash is improvement
# coefficient in comparison to scalar keccak1600-x86_64.pl.
# Digits in variables' names denote right-most coordinates:
my ($A00, # [0][0] [0][0] [0][0] [0][0] # %ymm0
$A01, # [0][4] [0][3] [0][2] [0][1] # %ymm1
$A20, # [3][0] [1][0] [4][0] [2][0] # %ymm2
$A31, # [2][4] [4][3] [1][2] [3][1] # %ymm3
$A21, # [3][4] [1][3] [4][2] [2][1] # %ymm4
$A41, # [1][4] [2][3] [3][2] [4][1] # %ymm5
$A11) = # [4][4] [3][3] [2][2] [1][1] # %ymm6
map("%ymm$_",(0..6));
# We also need to map the magic order into offsets within structure:
my @A_jagged = ([0,0], [1,0], [1,1], [1,2], [1,3], # [0][0..4]
[2,2], [6,0], [3,1], [4,2], [5,3], # [1][0..4]
[2,0], [4,0], [6,1], [5,2], [3,3], # [2][0..4]
[2,3], [3,0], [5,1], [6,2], [4,3], # [3][0..4]
[2,1], [5,0], [4,1], [3,2], [6,3]); # [4][0..4]
@A_jagged = map(8*($$_[0]*4+$$_[1]), @A_jagged); # ... and now linear
my @T = map("%ymm$_",(7..15));
my ($C14,$C00,$D00,$D14) = @T[5..8];
my ($R20,$R01,$R31,$R21,$R41,$R11) = map("%ymm$_",(16..21));
$code.=<<___;
.text
.type __KeccakF1600,\@function
.align 32
__KeccakF1600:
lea iotas(%rip),%r10
mov \$24,%eax
jmp .Loop_avx512vl
.align 32
.Loop_avx512vl:
######################################### Theta
vpshufd \$0b01001110,$A20,$C00
vpxor $A31,$A41,$C14
vpxor $A11,$A21,@T[2]
vpternlogq \$0x96,$A01,$T[2],$C14 # C[1..4]
vpxor $A20,$C00,$C00
vpermq \$0b01001110,$C00,@T[0]
vpermq \$0b10010011,$C14,@T[4]
vprolq \$1,$C14,@T[1] # ROL64(C[1..4],1)
vpermq \$0b00111001,@T[1],$D14
vpxor @T[4],@T[1],$D00
vpermq \$0b00000000,$D00,$D00 # D[0..0] = ROL64(C[1],1) ^ C[4]
vpternlogq \$0x96,@T[0],$A00,$C00 # C[0..0]
vprolq \$1,$C00,@T[1] # ROL64(C[0..0],1)
vpxor $D00,$A00,$A00 # ^= D[0..0]
vpblendd \$0b11000000,@T[1],$D14,$D14
vpblendd \$0b00000011,$C00,@T[4],@T[0]
######################################### Rho + Pi + pre-Chi shuffle
vpxor $D00,$A20,$A20 # ^= D[0..0] from Theta
vprolvq $R20,$A20,$A20
vpternlogq \$0x96,@T[0],$D14,$A31 # ^= D[1..4] from Theta
vprolvq $R31,$A31,$A31
vpternlogq \$0x96,@T[0],$D14,$A21 # ^= D[1..4] from Theta
vprolvq $R21,$A21,$A21
vpternlogq \$0x96,@T[0],$D14,$A41 # ^= D[1..4] from Theta
vprolvq $R41,$A41,$A41
vpermq \$0b10001101,$A20,@T[3] # $A20 -> future $A31
vpermq \$0b10001101,$A31,@T[4] # $A31 -> future $A21
vpternlogq \$0x96,@T[0],$D14,$A11 # ^= D[1..4] from Theta
vprolvq $R11,$A11,@T[1] # $A11 -> future $A01
vpermq \$0b00011011,$A21,@T[5] # $A21 -> future $A41
vpermq \$0b01110010,$A41,@T[6] # $A41 -> future $A11
vpternlogq \$0x96,@T[0],$D14,$A01 # ^= D[1..4] from Theta
vprolvq $R01,$A01,@T[2] # $A01 -> future $A20
######################################### Chi
vpblendd \$0b00001100,@T[6],@T[2],$A31 # [4][4] [2][0]
vpblendd \$0b00001100,@T[2],@T[4],@T[8] # [4][0] [2][1]
vpblendd \$0b00001100,@T[4],@T[3],$A41 # [4][2] [2][4]
vpblendd \$0b00001100,@T[3],@T[2],@T[7] # [4][3] [2][0]
vpblendd \$0b00110000,@T[4],$A31,$A31 # [1][3] [4][4] [2][0]
vpblendd \$0b00110000,@T[5],@T[8],@T[8] # [1][4] [4][0] [2][1]
vpblendd \$0b00110000,@T[2],$A41,$A41 # [1][0] [4][2] [2][4]
vpblendd \$0b00110000,@T[6],@T[7],@T[7] # [1][1] [4][3] [2][0]
vpblendd \$0b11000000,@T[5],$A31,$A31 # [3][2] [1][3] [4][4] [2][0]
vpblendd \$0b11000000,@T[6],@T[8],@T[8] # [3][3] [1][4] [4][0] [2][1]
vpblendd \$0b11000000,@T[6],$A41,$A41 # [3][3] [1][0] [4][2] [2][4]
vpblendd \$0b11000000,@T[4],@T[7],@T[7] # [3][4] [1][1] [4][3] [2][0]
vpternlogq \$0xC6,@T[8],@T[3],$A31 # [3][1] [1][2] [4][3] [2][4]
vpternlogq \$0xC6,@T[7],@T[5],$A41 # [3][2] [1][4] [4][1] [2][3]
vpsrldq \$8,@T[1],@T[0]
vpandn @T[0],@T[1],@T[0] # tgting [0][0] [0][0] [0][0] [0][0]
vpblendd \$0b00001100,@T[2],@T[5],$A11 # [4][0] [2][3]
vpblendd \$0b00001100,@T[5],@T[3],@T[8] # [4][1] [2][4]
vpblendd \$0b00110000,@T[3],$A11,$A11 # [1][2] [4][0] [2][3]
vpblendd \$0b00110000,@T[4],@T[8],@T[8] # [1][3] [4][1] [2][4]
vpblendd \$0b11000000,@T[4],$A11,$A11 # [3][4] [1][2] [4][0] [2][3]
vpblendd \$0b11000000,@T[2],@T[8],@T[8] # [3][0] [1][3] [4][1] [2][4]
vpternlogq \$0xC6,@T[8],@T[6],$A11 # [3][3] [1][1] [4][4] [2][2]
vpermq \$0b00011110,@T[1],$A21 # [0][1] [0][2] [0][4] [0][3]
vpblendd \$0b00110000,$A00,$A21,@T[8] # [0][1] [0][0] [0][4] [0][3]
vpermq \$0b00111001,@T[1],$A01 # [0][1] [0][4] [0][3] [0][2]
vpblendd \$0b11000000,$A00,$A01,$A01 # [0][0] [0][4] [0][3] [0][2]
vpblendd \$0b00001100,@T[5],@T[4],$A20 # [4][1] [2][1]
vpblendd \$0b00001100,@T[4],@T[6],@T[7] # [4][2] [2][2]
vpblendd \$0b00110000,@T[6],$A20,$A20 # [1][1] [4][1] [2][1]
vpblendd \$0b00110000,@T[3],@T[7],@T[7] # [1][2] [4][2] [2][2]
vpblendd \$0b11000000,@T[3],$A20,$A20 # [3][1] [1][1] [4][1] [2][1]
vpblendd \$0b11000000,@T[5],@T[7],@T[7] # [3][2] [1][2] [4][2] [2][2]
vpternlogq \$0xC6,@T[7],@T[2],$A20 # [3][0] [1][0] [4][0] [2][0]
vpermq \$0b00000000,@T[0],@T[0] # [0][0] [0][0] [0][0] [0][0]
vpermq \$0b00011011,$A31,$A31 # post-Chi shuffle
vpermq \$0b10001101,$A41,$A41
vpermq \$0b01110010,$A11,$A11
vpblendd \$0b00001100,@T[3],@T[6],$A21 # [4][3] [2][2]
vpblendd \$0b00001100,@T[6],@T[5],@T[7] # [4][4] [2][3]
vpblendd \$0b00110000,@T[5],$A21,$A21 # [1][4] [4][3] [2][2]
vpblendd \$0b00110000,@T[2],@T[7],@T[7] # [1][0] [4][4] [2][3]
vpblendd \$0b11000000,@T[2],$A21,$A21 # [3][0] [1][4] [4][3] [2][2]
vpblendd \$0b11000000,@T[3],@T[7],@T[7] # [3][1] [1][0] [4][4] [2][3]
vpternlogq \$0xC6,@T[8],@T[1],$A01 # [0][4] [0][3] [0][2] [0][1]
vpternlogq \$0xC6,@T[7],@T[4],$A21 # [3][4] [1][3] [4][2] [2][1]
######################################### Iota
vpternlogq \$0x96,(%r10),@T[0],$A00
lea 32(%r10),%r10
dec %eax
jnz .Loop_avx512vl
ret
.size __KeccakF1600,.-__KeccakF1600
___
my ($A_flat,$inp,$len,$bsz) = ("%rdi","%rsi","%rdx","%rcx");
my $out = $inp; # in squeeze
$code.=<<___;
.globl SHA3_absorb
.type SHA3_absorb,\@function
.align 32
SHA3_absorb:
mov %rsp,%r11
lea -240(%rsp),%rsp
|
