File: simon128_simd.cpp

package info (click to toggle)
libcrypto++ 8.2.0-2
  • links: PTS
  • area: main
  • in suites: experimental
  • size: 22,780 kB
  • sloc: cpp: 105,015; sh: 7,353; asm: 5,120; makefile: 316
file content (834 lines) | stat: -rw-r--r-- 29,944 bytes parent folder | download | duplicates (2)
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
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
// simon-simd.cpp - written and placed in the public domain by Jeffrey Walton
//
//    This source file uses intrinsics and built-ins to gain access to
//    SSSE3, ARM NEON and ARMv8a, and Power7 Altivec instructions. A separate
//    source file is needed because additional CXXFLAGS are required to enable
//    the appropriate instructions sets in some build configurations.

#include "pch.h"
#include "config.h"

#include "simon.h"
#include "misc.h"

// Uncomment for benchmarking C++ against SSE or NEON.
// Do so in both simon.cpp and simon-simd.cpp.
// #undef CRYPTOPP_SSSE3_AVAILABLE
// #undef CRYPTOPP_ARM_NEON_AVAILABLE

#if (CRYPTOPP_SSSE3_AVAILABLE)
# include "adv_simd.h"
# include <pmmintrin.h>
# include <tmmintrin.h>
#endif

#if defined(__XOP__)
# include <ammintrin.h>
#endif

#if defined(__AVX512F__)
# define CRYPTOPP_AVX512_ROTATE 1
# include <immintrin.h>
#endif

// C1189: error: This header is specific to ARM targets
#if (CRYPTOPP_ARM_NEON_AVAILABLE)
# include "adv_simd.h"
# ifndef _M_ARM64
#  include <arm_neon.h>
# endif
#endif

#if (CRYPTOPP_ARM_ACLE_AVAILABLE)
# include <stdint.h>
# include <arm_acle.h>
#endif

#if defined(CRYPTOPP_POWER8_AVAILABLE)
# include "adv_simd.h"
# include "ppc_simd.h"
#endif

// Squash MS LNK4221 and libtool warnings
extern const char SIMON128_SIMD_FNAME[] = __FILE__;

ANONYMOUS_NAMESPACE_BEGIN

using CryptoPP::byte;
using CryptoPP::word32;
using CryptoPP::word64;
using CryptoPP::vec_swap;  // SunCC

// *************************** ARM NEON ************************** //

#if (CRYPTOPP_ARM_NEON_AVAILABLE)

// Missing from Microsoft's ARM A-32 implementation
#if defined(_MSC_VER) && !defined(_M_ARM64)
inline uint64x2_t vld1q_dup_u64(const uint64_t* ptr)
{
	return vmovq_n_u64(*ptr);
}
#endif

template <class T>
inline T UnpackHigh64(const T& a, const T& b)
{
    const uint64x1_t x(vget_high_u64((uint64x2_t)a));
    const uint64x1_t y(vget_high_u64((uint64x2_t)b));
    return (T)vcombine_u64(x, y);
}

template <class T>
inline T UnpackLow64(const T& a, const T& b)
{
    const uint64x1_t x(vget_low_u64((uint64x2_t)a));
    const uint64x1_t y(vget_low_u64((uint64x2_t)b));
    return (T)vcombine_u64(x, y);
}

template <unsigned int R>
inline uint64x2_t RotateLeft64(const uint64x2_t& val)
{
    const uint64x2_t a(vshlq_n_u64(val, R));
    const uint64x2_t b(vshrq_n_u64(val, 64 - R));
    return vorrq_u64(a, b);
}

template <unsigned int R>
inline uint64x2_t RotateRight64(const uint64x2_t& val)
{
    const uint64x2_t a(vshlq_n_u64(val, 64 - R));
    const uint64x2_t b(vshrq_n_u64(val, R));
    return vorrq_u64(a, b);
}

#if defined(__aarch32__) || defined(__aarch64__)
// Faster than two Shifts and an Or. Thanks to Louis Wingers and Bryan Weeks.
template <>
inline uint64x2_t RotateLeft64<8>(const uint64x2_t& val)
{
    const uint8_t maskb[16] = { 7,0,1,2, 3,4,5,6, 15,8,9,10, 11,12,13,14 };
    const uint8x16_t mask = vld1q_u8(maskb);

    return vreinterpretq_u64_u8(
        vqtbl1q_u8(vreinterpretq_u8_u64(val), mask));
}

// Faster than two Shifts and an Or. Thanks to Louis Wingers and Bryan Weeks.
template <>
inline uint64x2_t RotateRight64<8>(const uint64x2_t& val)
{
    const uint8_t maskb[16] = { 1,2,3,4, 5,6,7,0, 9,10,11,12, 13,14,15,8 };
    const uint8x16_t mask = vld1q_u8(maskb);

    return vreinterpretq_u64_u8(
        vqtbl1q_u8(vreinterpretq_u8_u64(val), mask));
}
#endif

inline uint64x2_t SIMON128_f(const uint64x2_t& val)
{
    return veorq_u64(RotateLeft64<2>(val),
        vandq_u64(RotateLeft64<1>(val), RotateLeft64<8>(val)));
}

inline void SIMON128_Enc_Block(uint64x2_t &block0, uint64x2_t &block1,
    const word64 *subkeys, unsigned int rounds)
{
    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    uint64x2_t x1 = UnpackHigh64(block0, block1);
    uint64x2_t y1 = UnpackLow64(block0, block1);

    for (int i = 0; i < static_cast<int>(rounds & ~1)-1; i += 2)
    {
        const uint64x2_t rk1 = vld1q_dup_u64(subkeys+i);
        y1 = veorq_u64(veorq_u64(y1, SIMON128_f(x1)), rk1);

        const uint64x2_t rk2 = vld1q_dup_u64(subkeys+i+1);
        x1 = veorq_u64(veorq_u64(x1, SIMON128_f(y1)), rk2);
    }

    if (rounds & 1)
    {
        const uint64x2_t rk = vld1q_dup_u64(subkeys+rounds-1);

        y1 = veorq_u64(veorq_u64(y1, SIMON128_f(x1)), rk);
        std::swap(x1, y1);
    }

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block0 = UnpackLow64(y1, x1);
    block1 = UnpackHigh64(y1, x1);
}

inline void SIMON128_Enc_6_Blocks(uint64x2_t &block0, uint64x2_t &block1,
    uint64x2_t &block2, uint64x2_t &block3, uint64x2_t &block4, uint64x2_t &block5,
    const word64 *subkeys, unsigned int rounds)
{
    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    uint64x2_t x1 = UnpackHigh64(block0, block1);
    uint64x2_t y1 = UnpackLow64(block0, block1);
    uint64x2_t x2 = UnpackHigh64(block2, block3);
    uint64x2_t y2 = UnpackLow64(block2, block3);
    uint64x2_t x3 = UnpackHigh64(block4, block5);
    uint64x2_t y3 = UnpackLow64(block4, block5);

    for (int i = 0; i < static_cast<int>(rounds & ~1) - 1; i += 2)
    {
        const uint64x2_t rk1 = vld1q_dup_u64(subkeys+i);
        y1 = veorq_u64(veorq_u64(y1, SIMON128_f(x1)), rk1);
        y2 = veorq_u64(veorq_u64(y2, SIMON128_f(x2)), rk1);
        y3 = veorq_u64(veorq_u64(y3, SIMON128_f(x3)), rk1);

        const uint64x2_t rk2 = vld1q_dup_u64(subkeys+i+1);
        x1 = veorq_u64(veorq_u64(x1, SIMON128_f(y1)), rk2);
        x2 = veorq_u64(veorq_u64(x2, SIMON128_f(y2)), rk2);
        x3 = veorq_u64(veorq_u64(x3, SIMON128_f(y3)), rk2);
    }

    if (rounds & 1)
    {
        const uint64x2_t rk = vld1q_dup_u64(subkeys + rounds - 1);

        y1 = veorq_u64(veorq_u64(y1, SIMON128_f(x1)), rk);
        y2 = veorq_u64(veorq_u64(y2, SIMON128_f(x2)), rk);
        y3 = veorq_u64(veorq_u64(y3, SIMON128_f(x3)), rk);
        std::swap(x1, y1); std::swap(x2, y2); std::swap(x3, y3);
    }

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block0 = UnpackLow64(y1, x1);
    block1 = UnpackHigh64(y1, x1);
    block2 = UnpackLow64(y2, x2);
    block3 = UnpackHigh64(y2, x2);
    block4 = UnpackLow64(y3, x3);
    block5 = UnpackHigh64(y3, x3);
}

inline void SIMON128_Dec_Block(uint64x2_t &block0, uint64x2_t &block1,
    const word64 *subkeys, unsigned int rounds)
{
    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    uint64x2_t x1 = UnpackHigh64(block0, block1);
    uint64x2_t y1 = UnpackLow64(block0, block1);

    if (rounds & 1)
    {
        std::swap(x1, y1);
        const uint64x2_t rk = vld1q_dup_u64(subkeys + rounds - 1);

        y1 = veorq_u64(veorq_u64(y1, rk), SIMON128_f(x1));
        rounds--;
    }

    for (int i = static_cast<int>(rounds-2); i >= 0; i -= 2)
    {
        const uint64x2_t rk1 = vld1q_dup_u64(subkeys+i+1);
        x1 = veorq_u64(veorq_u64(x1, SIMON128_f(y1)), rk1);

        const uint64x2_t rk2 = vld1q_dup_u64(subkeys+i);
        y1 = veorq_u64(veorq_u64(y1, SIMON128_f(x1)), rk2);
    }

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block0 = UnpackLow64(y1, x1);
    block1 = UnpackHigh64(y1, x1);
}

inline void SIMON128_Dec_6_Blocks(uint64x2_t &block0, uint64x2_t &block1,
    uint64x2_t &block2, uint64x2_t &block3, uint64x2_t &block4, uint64x2_t &block5,
    const word64 *subkeys, unsigned int rounds)
{
    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    uint64x2_t x1 = UnpackHigh64(block0, block1);
    uint64x2_t y1 = UnpackLow64(block0, block1);
    uint64x2_t x2 = UnpackHigh64(block2, block3);
    uint64x2_t y2 = UnpackLow64(block2, block3);
    uint64x2_t x3 = UnpackHigh64(block4, block5);
    uint64x2_t y3 = UnpackLow64(block4, block5);

    if (rounds & 1)
    {
        std::swap(x1, y1); std::swap(x2, y2); std::swap(x3, y3);
        const uint64x2_t rk = vld1q_dup_u64(subkeys + rounds - 1);

        y1 = veorq_u64(veorq_u64(y1, rk), SIMON128_f(x1));
        y2 = veorq_u64(veorq_u64(y2, rk), SIMON128_f(x2));
        y3 = veorq_u64(veorq_u64(y3, rk), SIMON128_f(x3));
        rounds--;
    }

    for (int i = static_cast<int>(rounds-2); i >= 0; i -= 2)
    {
        const uint64x2_t rk1 = vld1q_dup_u64(subkeys + i + 1);
        x1 = veorq_u64(veorq_u64(x1, SIMON128_f(y1)), rk1);
        x2 = veorq_u64(veorq_u64(x2, SIMON128_f(y2)), rk1);
        x3 = veorq_u64(veorq_u64(x3, SIMON128_f(y3)), rk1);

        const uint64x2_t rk2 = vld1q_dup_u64(subkeys + i);
        y1 = veorq_u64(veorq_u64(y1, SIMON128_f(x1)), rk2);
        y2 = veorq_u64(veorq_u64(y2, SIMON128_f(x2)), rk2);
        y3 = veorq_u64(veorq_u64(y3, SIMON128_f(x3)), rk2);
    }

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block0 = UnpackLow64(y1, x1);
    block1 = UnpackHigh64(y1, x1);
    block2 = UnpackLow64(y2, x2);
    block3 = UnpackHigh64(y2, x2);
    block4 = UnpackLow64(y3, x3);
    block5 = UnpackHigh64(y3, x3);
}

#endif  // CRYPTOPP_ARM_NEON_AVAILABLE

// ***************************** IA-32 ***************************** //

#if defined(CRYPTOPP_SSSE3_AVAILABLE)

// Clang __m128i casts, http://bugs.llvm.org/show_bug.cgi?id=20670
#ifndef M128_CAST
# define M128_CAST(x) ((__m128i *)(void *)(x))
#endif
#ifndef CONST_M128_CAST
# define CONST_M128_CAST(x) ((const __m128i *)(const void *)(x))
#endif

// GCC double casts, https://www.spinics.net/lists/gcchelp/msg47735.html
#ifndef DOUBLE_CAST
# define DOUBLE_CAST(x) ((double *)(void *)(x))
#endif
#ifndef CONST_DOUBLE_CAST
# define CONST_DOUBLE_CAST(x) ((const double *)(const void *)(x))
#endif

inline void Swap128(__m128i& a,__m128i& b)
{
#if defined(__SUNPRO_CC) && (__SUNPRO_CC <= 0x5120)
    // __m128i is an unsigned long long[2], and support for swapping it was not added until C++11.
    // SunCC 12.1 - 12.3 fail to consume the swap; while SunCC 12.4 consumes it without -std=c++11.
    vec_swap(a, b);
#else
    std::swap(a, b);
#endif
}

template <unsigned int R>
inline __m128i RotateLeft64(const __m128i& val)
{
#if defined(CRYPTOPP_AVX512_ROTATE)
    return _mm_rol_epi64(val, R);
#elif defined(__XOP__)
    return _mm_roti_epi64(val, R);
#else
    return _mm_or_si128(
        _mm_slli_epi64(val, R), _mm_srli_epi64(val, 64-R));
#endif
}

template <unsigned int R>
inline __m128i RotateRight64(const __m128i& val)
{
#if defined(CRYPTOPP_AVX512_ROTATE)
    return _mm_ror_epi64(val, R);
#elif defined(__XOP__)
    return _mm_roti_epi64(val, 64-R);
#else
    return _mm_or_si128(
        _mm_slli_epi64(val, 64-R), _mm_srli_epi64(val, R));
#endif
}

// Faster than two Shifts and an Or. Thanks to Louis Wingers and Bryan Weeks.
template <>
__m128i RotateLeft64<8>(const __m128i& val)
{
#if defined(__XOP__)
    return _mm_roti_epi64(val, 8);
#else
    const __m128i mask = _mm_set_epi8(14,13,12,11, 10,9,8,15, 6,5,4,3, 2,1,0,7);
    return _mm_shuffle_epi8(val, mask);
#endif
}

// Faster than two Shifts and an Or. Thanks to Louis Wingers and Bryan Weeks.
template <>
__m128i RotateRight64<8>(const __m128i& val)
{
#if defined(__XOP__)
    return _mm_roti_epi64(val, 64-8);
#else
    const __m128i mask = _mm_set_epi8(8,15,14,13, 12,11,10,9, 0,7,6,5, 4,3,2,1);
    return _mm_shuffle_epi8(val, mask);
#endif
}

inline __m128i SIMON128_f(const __m128i& v)
{
    return _mm_xor_si128(RotateLeft64<2>(v),
        _mm_and_si128(RotateLeft64<1>(v), RotateLeft64<8>(v)));
}

inline void SIMON128_Enc_Block(__m128i &block0, __m128i &block1,
    const word64 *subkeys, unsigned int rounds)
{
    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    __m128i x1 = _mm_unpackhi_epi64(block0, block1);
    __m128i y1 = _mm_unpacklo_epi64(block0, block1);

    for (int i = 0; i < static_cast<int>(rounds & ~1)-1; i += 2)
    {
        const __m128i rk1 = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys+i)));
        y1 = _mm_xor_si128(_mm_xor_si128(y1, SIMON128_f(x1)), rk1);

        const __m128i rk2 = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys+i+1)));
        x1 = _mm_xor_si128(_mm_xor_si128(x1, SIMON128_f(y1)), rk2);
    }

    if (rounds & 1)
    {
        const __m128i rk = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys+rounds-1)));

        y1 = _mm_xor_si128(_mm_xor_si128(y1, SIMON128_f(x1)), rk);
        Swap128(x1, y1);
    }

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block0 = _mm_unpacklo_epi64(y1, x1);
    block1 = _mm_unpackhi_epi64(y1, x1);
}

inline void SIMON128_Enc_6_Blocks(__m128i &block0, __m128i &block1,
    __m128i &block2, __m128i &block3, __m128i &block4, __m128i &block5,
    const word64 *subkeys, unsigned int rounds)
{
    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    __m128i x1 = _mm_unpackhi_epi64(block0, block1);
    __m128i y1 = _mm_unpacklo_epi64(block0, block1);
    __m128i x2 = _mm_unpackhi_epi64(block2, block3);
    __m128i y2 = _mm_unpacklo_epi64(block2, block3);
    __m128i x3 = _mm_unpackhi_epi64(block4, block5);
    __m128i y3 = _mm_unpacklo_epi64(block4, block5);

    for (int i = 0; i < static_cast<int>(rounds & ~1) - 1; i += 2)
    {
        const __m128i rk1 = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys + i)));
        y1 = _mm_xor_si128(_mm_xor_si128(y1, SIMON128_f(x1)), rk1);
        y2 = _mm_xor_si128(_mm_xor_si128(y2, SIMON128_f(x2)), rk1);
        y3 = _mm_xor_si128(_mm_xor_si128(y3, SIMON128_f(x3)), rk1);

        const __m128i rk2 = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys + i + 1)));
        x1 = _mm_xor_si128(_mm_xor_si128(x1, SIMON128_f(y1)), rk2);
        x2 = _mm_xor_si128(_mm_xor_si128(x2, SIMON128_f(y2)), rk2);
        x3 = _mm_xor_si128(_mm_xor_si128(x3, SIMON128_f(y3)), rk2);
    }

    if (rounds & 1)
    {
        const __m128i rk = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys + rounds - 1)));
        y1 = _mm_xor_si128(_mm_xor_si128(y1, SIMON128_f(x1)), rk);
        y2 = _mm_xor_si128(_mm_xor_si128(y2, SIMON128_f(x2)), rk);
        y3 = _mm_xor_si128(_mm_xor_si128(y3, SIMON128_f(x3)), rk);
        Swap128(x1, y1); Swap128(x2, y2); Swap128(x3, y3);
    }

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block0 = _mm_unpacklo_epi64(y1, x1);
    block1 = _mm_unpackhi_epi64(y1, x1);
    block2 = _mm_unpacklo_epi64(y2, x2);
    block3 = _mm_unpackhi_epi64(y2, x2);
    block4 = _mm_unpacklo_epi64(y3, x3);
    block5 = _mm_unpackhi_epi64(y3, x3);
}

inline void SIMON128_Dec_Block(__m128i &block0, __m128i &block1,
    const word64 *subkeys, unsigned int rounds)
{
    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    __m128i x1 = _mm_unpackhi_epi64(block0, block1);
    __m128i y1 = _mm_unpacklo_epi64(block0, block1);

    if (rounds & 1)
    {
        const __m128i rk = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys + rounds - 1)));

        Swap128(x1, y1);
        y1 = _mm_xor_si128(_mm_xor_si128(y1, rk), SIMON128_f(x1));
        rounds--;
    }

    for (int i = static_cast<int>(rounds-2); i >= 0; i -= 2)
    {
        const __m128i rk1 = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys+i+1)));
        x1 = _mm_xor_si128(_mm_xor_si128(x1, SIMON128_f(y1)), rk1);

        const __m128i rk2 = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys+i)));
        y1 = _mm_xor_si128(_mm_xor_si128(y1, SIMON128_f(x1)), rk2);
    }

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block0 = _mm_unpacklo_epi64(y1, x1);
    block1 = _mm_unpackhi_epi64(y1, x1);
}

inline void SIMON128_Dec_6_Blocks(__m128i &block0, __m128i &block1,
    __m128i &block2, __m128i &block3, __m128i &block4, __m128i &block5,
    const word64 *subkeys, unsigned int rounds)
{
    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    __m128i x1 = _mm_unpackhi_epi64(block0, block1);
    __m128i y1 = _mm_unpacklo_epi64(block0, block1);
    __m128i x2 = _mm_unpackhi_epi64(block2, block3);
    __m128i y2 = _mm_unpacklo_epi64(block2, block3);
    __m128i x3 = _mm_unpackhi_epi64(block4, block5);
    __m128i y3 = _mm_unpacklo_epi64(block4, block5);

    if (rounds & 1)
    {
        const __m128i rk = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys + rounds - 1)));

        Swap128(x1, y1); Swap128(x2, y2); Swap128(x3, y3);
        y1 = _mm_xor_si128(_mm_xor_si128(y1, rk), SIMON128_f(x1));
        y2 = _mm_xor_si128(_mm_xor_si128(y2, rk), SIMON128_f(x2));
        y3 = _mm_xor_si128(_mm_xor_si128(y3, rk), SIMON128_f(x3));
        rounds--;
    }

    for (int i = static_cast<int>(rounds-2); i >= 0; i -= 2)
    {
        const __m128i rk1 = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys + i + 1)));
        x1 = _mm_xor_si128(_mm_xor_si128(x1, SIMON128_f(y1)), rk1);
        x2 = _mm_xor_si128(_mm_xor_si128(x2, SIMON128_f(y2)), rk1);
        x3 = _mm_xor_si128(_mm_xor_si128(x3, SIMON128_f(y3)), rk1);

        const __m128i rk2 = _mm_castpd_si128(
            _mm_loaddup_pd(CONST_DOUBLE_CAST(subkeys + i)));
        y1 = _mm_xor_si128(_mm_xor_si128(y1, SIMON128_f(x1)), rk2);
        y2 = _mm_xor_si128(_mm_xor_si128(y2, SIMON128_f(x2)), rk2);
        y3 = _mm_xor_si128(_mm_xor_si128(y3, SIMON128_f(x3)), rk2);
    }

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block0 = _mm_unpacklo_epi64(y1, x1);
    block1 = _mm_unpackhi_epi64(y1, x1);
    block2 = _mm_unpacklo_epi64(y2, x2);
    block3 = _mm_unpackhi_epi64(y2, x2);
    block4 = _mm_unpacklo_epi64(y3, x3);
    block5 = _mm_unpackhi_epi64(y3, x3);
}

#endif  // CRYPTOPP_SSSE3_AVAILABLE

// ***************************** Power8 ***************************** //

#if defined(CRYPTOPP_POWER8_AVAILABLE)

using CryptoPP::uint8x16_p;
using CryptoPP::uint32x4_p;
using CryptoPP::uint64x2_p;

using CryptoPP::VecAnd;
using CryptoPP::VecXor;
using CryptoPP::VecPermute;

// Rotate left by bit count
template<unsigned int C>
inline uint64x2_p RotateLeft64(const uint64x2_p val)
{
    const uint64x2_p m = {C, C};
    return vec_rl(val, m);
}

// Rotate right by bit count
template<unsigned int C>
inline uint64x2_p RotateRight64(const uint64x2_p val)
{
    const uint64x2_p m = {64-C, 64-C};
    return vec_rl(val, m);
}

inline uint64x2_p SIMON128_f(const uint64x2_p val)
{
    return VecXor(RotateLeft64<2>(val),
        VecAnd(RotateLeft64<1>(val), RotateLeft64<8>(val)));
}

inline void SIMON128_Enc_Block(uint32x4_p &block, const word64 *subkeys, unsigned int rounds)
{
#if (CRYPTOPP_BIG_ENDIAN)
    const uint8x16_p m1 = {31,30,29,28,27,26,25,24, 15,14,13,12,11,10,9,8};
    const uint8x16_p m2 = {23,22,21,20,19,18,17,16, 7,6,5,4,3,2,1,0};
#else
    const uint8x16_p m1 = {7,6,5,4,3,2,1,0, 23,22,21,20,19,18,17,16};
    const uint8x16_p m2 = {15,14,13,12,11,10,9,8, 31,30,29,28,27,26,25,24};
#endif

    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    uint64x2_p x1 = (uint64x2_p)VecPermute(block, block, m1);
    uint64x2_p y1 = (uint64x2_p)VecPermute(block, block, m2);

    for (int i = 0; i < static_cast<int>(rounds & ~1)-1; i += 2)
    {
        const uint64x2_p rk1 = vec_splats((unsigned long long)subkeys[i]);
        y1 = VecXor(VecXor(y1, SIMON128_f(x1)), rk1);

        const uint64x2_p rk2 = vec_splats((unsigned long long)subkeys[i+1]);
        x1 = VecXor(VecXor(x1, SIMON128_f(y1)), rk2);
    }

    if (rounds & 1)
    {
        const uint64x2_p rk = vec_splats((unsigned long long)subkeys[rounds-1]);
        y1 = VecXor(VecXor(y1, SIMON128_f(x1)), rk);
        std::swap(x1, y1);
    }

#if (CRYPTOPP_BIG_ENDIAN)
    const uint8x16_p m3 = {31,30,29,28,27,26,25,24, 15,14,13,12,11,10,9,8};
    //const uint8x16_p m4 = {23,22,21,20,19,18,17,16, 7,6,5,4,3,2,1,0};
#else
    const uint8x16_p m3 = {7,6,5,4,3,2,1,0, 23,22,21,20,19,18,17,16};
    //const uint8x16_p m4 = {15,14,13,12,11,10,9,8, 31,30,29,28,27,26,25,24};
#endif

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block = (uint32x4_p)VecPermute(x1, y1, m3);
}

inline void SIMON128_Dec_Block(uint32x4_p &block, const word64 *subkeys, unsigned int rounds)
{
#if (CRYPTOPP_BIG_ENDIAN)
    const uint8x16_p m1 = {31,30,29,28,27,26,25,24, 15,14,13,12,11,10,9,8};
    const uint8x16_p m2 = {23,22,21,20,19,18,17,16, 7,6,5,4,3,2,1,0};
#else
    const uint8x16_p m1 = {7,6,5,4,3,2,1,0, 23,22,21,20,19,18,17,16};
    const uint8x16_p m2 = {15,14,13,12,11,10,9,8, 31,30,29,28,27,26,25,24};
#endif

    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    uint64x2_p x1 = (uint64x2_p)VecPermute(block, block, m1);
    uint64x2_p y1 = (uint64x2_p)VecPermute(block, block, m2);

    if (rounds & 1)
    {
        std::swap(x1, y1);
        const uint64x2_p rk = vec_splats((unsigned long long)subkeys[rounds-1]);
        y1 = VecXor(VecXor(y1, rk), SIMON128_f(x1));
        rounds--;
    }

    for (int i = static_cast<int>(rounds-2); i >= 0; i -= 2)
    {
        const uint64x2_p rk1 = vec_splats((unsigned long long)subkeys[i+1]);
        x1 = VecXor(VecXor(x1, SIMON128_f(y1)), rk1);

        const uint64x2_p rk2 = vec_splats((unsigned long long)subkeys[i]);
        y1 = VecXor(VecXor(y1, SIMON128_f(x1)), rk2);
    }

#if (CRYPTOPP_BIG_ENDIAN)
    const uint8x16_p m3 = {31,30,29,28,27,26,25,24, 15,14,13,12,11,10,9,8};
    //const uint8x16_p m4 = {23,22,21,20,19,18,17,16, 7,6,5,4,3,2,1,0};
#else
    const uint8x16_p m3 = {7,6,5,4,3,2,1,0, 23,22,21,20,19,18,17,16};
    //const uint8x16_p m4 = {15,14,13,12,11,10,9,8, 31,30,29,28,27,26,25,24};
#endif

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block = (uint32x4_p)VecPermute(x1, y1, m3);
}

inline void SIMON128_Enc_6_Blocks(uint32x4_p &block0, uint32x4_p &block1,
            uint32x4_p &block2, uint32x4_p &block3, uint32x4_p &block4,
            uint32x4_p &block5, const word64 *subkeys, unsigned int rounds)
{
#if (CRYPTOPP_BIG_ENDIAN)
    const uint8x16_p m1 = {31,30,29,28,27,26,25,24, 15,14,13,12,11,10,9,8};
    const uint8x16_p m2 = {23,22,21,20,19,18,17,16, 7,6,5,4,3,2,1,0};
#else
    const uint8x16_p m1 = {7,6,5,4,3,2,1,0, 23,22,21,20,19,18,17,16};
    const uint8x16_p m2 = {15,14,13,12,11,10,9,8, 31,30,29,28,27,26,25,24};
#endif

    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    uint64x2_p x1 = (uint64x2_p)VecPermute(block0, block1, m1);
    uint64x2_p y1 = (uint64x2_p)VecPermute(block0, block1, m2);
    uint64x2_p x2 = (uint64x2_p)VecPermute(block2, block3, m1);
    uint64x2_p y2 = (uint64x2_p)VecPermute(block2, block3, m2);
    uint64x2_p x3 = (uint64x2_p)VecPermute(block4, block5, m1);
    uint64x2_p y3 = (uint64x2_p)VecPermute(block4, block5, m2);

    for (int i = 0; i < static_cast<int>(rounds & ~1)-1; i += 2)
    {
        const uint64x2_p rk1 = vec_splats((unsigned long long)subkeys[i]);
        y1 = VecXor(VecXor(y1, SIMON128_f(x1)), rk1);
        y2 = VecXor(VecXor(y2, SIMON128_f(x2)), rk1);
        y3 = VecXor(VecXor(y3, SIMON128_f(x3)), rk1);

        const uint64x2_p rk2 = vec_splats((unsigned long long)subkeys[i+1]);
        x1 = VecXor(VecXor(x1, SIMON128_f(y1)), rk2);
        x2 = VecXor(VecXor(x2, SIMON128_f(y2)), rk2);
        x3 = VecXor(VecXor(x3, SIMON128_f(y3)), rk2);
    }

    if (rounds & 1)
    {
        const uint64x2_p rk = vec_splats((unsigned long long)subkeys[rounds-1]);
        y1 = VecXor(VecXor(y1, SIMON128_f(x1)), rk);
        y2 = VecXor(VecXor(y2, SIMON128_f(x2)), rk);
        y3 = VecXor(VecXor(y3, SIMON128_f(x3)), rk);
        std::swap(x1, y1); std::swap(x2, y2); std::swap(x3, y3);
    }

#if (CRYPTOPP_BIG_ENDIAN)
    const uint8x16_p m3 = {31,30,29,28,27,26,25,24, 15,14,13,12,11,10,9,8};
    const uint8x16_p m4 = {23,22,21,20,19,18,17,16, 7,6,5,4,3,2,1,0};
#else
    const uint8x16_p m3 = {7,6,5,4,3,2,1,0, 23,22,21,20,19,18,17,16};
    const uint8x16_p m4 = {15,14,13,12,11,10,9,8, 31,30,29,28,27,26,25,24};
#endif

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block0 = (uint32x4_p)VecPermute(x1, y1, m3);
    block1 = (uint32x4_p)VecPermute(x1, y1, m4);
    block2 = (uint32x4_p)VecPermute(x2, y2, m3);
    block3 = (uint32x4_p)VecPermute(x2, y2, m4);
    block4 = (uint32x4_p)VecPermute(x3, y3, m3);
    block5 = (uint32x4_p)VecPermute(x3, y3, m4);
}

inline void SIMON128_Dec_6_Blocks(uint32x4_p &block0, uint32x4_p &block1,
            uint32x4_p &block2, uint32x4_p &block3, uint32x4_p &block4,
            uint32x4_p &block5, const word64 *subkeys, unsigned int rounds)
{
#if (CRYPTOPP_BIG_ENDIAN)
    const uint8x16_p m1 = {31,30,29,28,27,26,25,24, 15,14,13,12,11,10,9,8};
    const uint8x16_p m2 = {23,22,21,20,19,18,17,16, 7,6,5,4,3,2,1,0};
#else
    const uint8x16_p m1 = {7,6,5,4,3,2,1,0, 23,22,21,20,19,18,17,16};
    const uint8x16_p m2 = {15,14,13,12,11,10,9,8, 31,30,29,28,27,26,25,24};
#endif

    // [A1 A2][B1 B2] ... => [A1 B1][A2 B2] ...
    uint64x2_p x1 = (uint64x2_p)VecPermute(block0, block1, m1);
    uint64x2_p y1 = (uint64x2_p)VecPermute(block0, block1, m2);
    uint64x2_p x2 = (uint64x2_p)VecPermute(block2, block3, m1);
    uint64x2_p y2 = (uint64x2_p)VecPermute(block2, block3, m2);
    uint64x2_p x3 = (uint64x2_p)VecPermute(block4, block5, m1);
    uint64x2_p y3 = (uint64x2_p)VecPermute(block4, block5, m2);

    if (rounds & 1)
    {
        std::swap(x1, y1); std::swap(x2, y2); std::swap(x3, y3);
        const uint64x2_p rk = vec_splats((unsigned long long)subkeys[rounds-1]);
        y1 = VecXor(VecXor(y1, rk), SIMON128_f(x1));
        y2 = VecXor(VecXor(y2, rk), SIMON128_f(x2));
        y3 = VecXor(VecXor(y3, rk), SIMON128_f(x3));
        rounds--;
    }

    for (int i = static_cast<int>(rounds-2); i >= 0; i -= 2)
    {
        const uint64x2_p rk1 = vec_splats((unsigned long long)subkeys[i+1]);
        x1 = VecXor(VecXor(x1, SIMON128_f(y1)), rk1);
        x2 = VecXor(VecXor(x2, SIMON128_f(y2)), rk1);
        x3 = VecXor(VecXor(x3, SIMON128_f(y3)), rk1);

        const uint64x2_p rk2 = vec_splats((unsigned long long)subkeys[i]);
        y1 = VecXor(VecXor(y1, SIMON128_f(x1)), rk2);
        y2 = VecXor(VecXor(y2, SIMON128_f(x2)), rk2);
        y3 = VecXor(VecXor(y3, SIMON128_f(x3)), rk2);
    }

#if (CRYPTOPP_BIG_ENDIAN)
    const uint8x16_p m3 = {31,30,29,28,27,26,25,24, 15,14,13,12,11,10,9,8};
    const uint8x16_p m4 = {23,22,21,20,19,18,17,16, 7,6,5,4,3,2,1,0};
#else
    const uint8x16_p m3 = {7,6,5,4,3,2,1,0, 23,22,21,20,19,18,17,16};
    const uint8x16_p m4 = {15,14,13,12,11,10,9,8, 31,30,29,28,27,26,25,24};
#endif

    // [A1 B1][A2 B2] ... => [A1 A2][B1 B2] ...
    block0 = (uint32x4_p)VecPermute(x1, y1, m3);
    block1 = (uint32x4_p)VecPermute(x1, y1, m4);
    block2 = (uint32x4_p)VecPermute(x2, y2, m3);
    block3 = (uint32x4_p)VecPermute(x2, y2, m4);
    block4 = (uint32x4_p)VecPermute(x3, y3, m3);
    block5 = (uint32x4_p)VecPermute(x3, y3, m4);
}

#endif  // CRYPTOPP_POWER8_AVAILABLE

ANONYMOUS_NAMESPACE_END

///////////////////////////////////////////////////////////////////////

NAMESPACE_BEGIN(CryptoPP)

// *************************** ARM NEON **************************** //

#if (CRYPTOPP_ARM_NEON_AVAILABLE)
size_t SIMON128_Enc_AdvancedProcessBlocks_NEON(const word64* subKeys, size_t rounds,
    const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags)
{
    return AdvancedProcessBlocks128_6x2_NEON(SIMON128_Enc_Block, SIMON128_Enc_6_Blocks,
        subKeys, rounds, inBlocks, xorBlocks, outBlocks, length, flags);
}

size_t SIMON128_Dec_AdvancedProcessBlocks_NEON(const word64* subKeys, size_t rounds,
    const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags)
{
    return AdvancedProcessBlocks128_6x2_NEON(SIMON128_Dec_Block, SIMON128_Dec_6_Blocks,
        subKeys, rounds, inBlocks, xorBlocks, outBlocks, length, flags);
}
#endif  // CRYPTOPP_ARM_NEON_AVAILABLE

// ***************************** IA-32 ***************************** //

#if defined(CRYPTOPP_SSSE3_AVAILABLE)
size_t SIMON128_Enc_AdvancedProcessBlocks_SSSE3(const word64* subKeys, size_t rounds,
    const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags)
{
    return AdvancedProcessBlocks128_6x2_SSE(SIMON128_Enc_Block, SIMON128_Enc_6_Blocks,
        subKeys, rounds, inBlocks, xorBlocks, outBlocks, length, flags);
}

size_t SIMON128_Dec_AdvancedProcessBlocks_SSSE3(const word64* subKeys, size_t rounds,
    const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags)
{
    return AdvancedProcessBlocks128_6x2_SSE(SIMON128_Dec_Block, SIMON128_Dec_6_Blocks,
        subKeys, rounds, inBlocks, xorBlocks, outBlocks, length, flags);
}
#endif  // CRYPTOPP_SSSE3_AVAILABLE

// ***************************** Power8 ***************************** //

#if defined(CRYPTOPP_POWER8_AVAILABLE)
size_t SIMON128_Enc_AdvancedProcessBlocks_POWER8(const word64* subKeys, size_t rounds,
    const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags)
{
    return AdvancedProcessBlocks128_6x1_ALTIVEC(SIMON128_Enc_Block, SIMON128_Enc_6_Blocks,
        subKeys, rounds, inBlocks, xorBlocks, outBlocks, length, flags);
}

size_t SIMON128_Dec_AdvancedProcessBlocks_POWER8(const word64* subKeys, size_t rounds,
    const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags)
{
    return AdvancedProcessBlocks128_6x1_ALTIVEC(SIMON128_Dec_Block, SIMON128_Dec_6_Blocks,
        subKeys, rounds, inBlocks, xorBlocks, outBlocks, length, flags);
}
#endif  // CRYPTOPP_POWER8_AVAILABLE

NAMESPACE_END