popov/cpuminer-opt-gpu
1
0
Fork 0
mirror of https://github.com/JayDDee/cpuminer-opt.git synced 2025-09-17 23:44:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

v3.11.9

Browse Source
This commit is contained in:
Jay D Dee
2020-02-04 01:31:59 -05:00
parent 0681ca996d
commit 1b76cee239
106 changed files with 1695 additions and 4481 deletions
Download Patch File Download Diff File Expand all files Collapse all files

186
algo/groestl/aes_ni/groestl-intr-aes.h
View File

@@ -1,3 +1,6 @@
#if !defined GROESTL_INTR_AES_H__
#define GROESTL_INTR_AES_H__
/* groestl-intr-aes.h Aug 2011
*
* Groestl implementation with intrinsics using ssse3, sse4.1, and aes
@@ -11,6 +14,52 @@
#include <wmmintrin.h>
#include "hash-groestl.h"
static const __m128i round_const_p[] __attribute__ ((aligned (64))) =
{
{ 0x7060504030201000, 0xf0e0d0c0b0a09080 },
{ 0x7161514131211101, 0xf1e1d1c1b1a19181 },
{ 0x7262524232221202, 0xf2e2d2c2b2a29282 },
{ 0x7363534333231303, 0xf3e3d3c3b3a39383 },
{ 0x7464544434241404, 0xf4e4d4c4b4a49484 },
{ 0x7565554535251505, 0xf5e5d5c5b5a59585 },
{ 0x7666564636261606, 0xf6e6d6c6b6a69686 },
{ 0x7767574737271707, 0xf7e7d7c7b7a79787 },
{ 0x7868584838281808, 0xf8e8d8c8b8a89888 },
{ 0x7969594939291909, 0xf9e9d9c9b9a99989 },
{ 0x7a6a5a4a3a2a1a0a, 0xfaeadacabaaa9a8a },
{ 0x7b6b5b4b3b2b1b0b, 0xfbebdbcbbbab9b8b },
{ 0x7c6c5c4c3c2c1c0c, 0xfcecdcccbcac9c8c },
{ 0x7d6d5d4d3d2d1d0d, 0xfdedddcdbdad9d8d }
};
static const __m128i round_const_q[] __attribute__ ((aligned (64))) =
{
{ 0x8f9fafbfcfdfefff, 0x0f1f2f3f4f5f6f7f },
{ 0x8e9eaebecedeeefe, 0x0e1e2e3e4e5e6e7e },
{ 0x8d9dadbdcdddedfd, 0x0d1d2d3d4d5d6d7d },
{ 0x8c9cacbcccdcecfc, 0x0c1c2c3c4c5c6c7c },
{ 0x8b9babbbcbdbebfb, 0x0b1b2b3b4b5b6b7b },
{ 0x8a9aaabacadaeafa, 0x0a1a2a3a4a5a6a7a },
{ 0x8999a9b9c9d9e9f9, 0x0919293949596979 },
{ 0x8898a8b8c8d8e8f8, 0x0818283848586878 },
{ 0x8797a7b7c7d7e7f7, 0x0717273747576777 },
{ 0x8696a6b6c6d6e6f6, 0x0616263646566676 },
{ 0x8595a5b5c5d5e5f5, 0x0515253545556575 },
{ 0x8494a4b4c4d4e4f4, 0x0414243444546474 },
{ 0x8393a3b3c3d3e3f3, 0x0313233343536373 },
{ 0x8292a2b2c2d2e2f2, 0x0212223242526272 }
};
static const __m128i TRANSP_MASK = { 0x0d0509010c040800, 0x0f070b030e060a02 };
static const __m128i SUBSH_MASK0 = { 0x0b0e0104070a0d00, 0x0306090c0f020508 };
static const __m128i SUBSH_MASK1 = { 0x0c0f0205080b0e01, 0x04070a0d00030609 };
static const __m128i SUBSH_MASK2 = { 0x0d000306090c0f02, 0x05080b0e0104070a };
static const __m128i SUBSH_MASK3 = { 0x0e0104070a0d0003, 0x06090c0f0205080b };
static const __m128i SUBSH_MASK4 = { 0x0f0205080b0e0104, 0x070a0d000306090c };
static const __m128i SUBSH_MASK5 = { 0x000306090c0f0205, 0x080b0e0104070a0d };
static const __m128i SUBSH_MASK6 = { 0x0104070a0d000306, 0x090c0f0205080b0e };
static const __m128i SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003 };
#define tos(a) #a
#define tostr(a) tos(a)
@@ -141,42 +190,6 @@
}/*MixBytes*/
static const uint64_t round_const_p[] __attribute__ ((aligned (64))) =
{
0x7060504030201000, 0xf0e0d0c0b0a09080,
0x7161514131211101, 0xf1e1d1c1b1a19181,
0x7262524232221202, 0xf2e2d2c2b2a29282,
0x7363534333231303, 0xf3e3d3c3b3a39383,
0x7464544434241404, 0xf4e4d4c4b4a49484,
0x7565554535251505, 0xf5e5d5c5b5a59585,
0x7666564636261606, 0xf6e6d6c6b6a69686,
0x7767574737271707, 0xf7e7d7c7b7a79787,
0x7868584838281808, 0xf8e8d8c8b8a89888,
0x7969594939291909, 0xf9e9d9c9b9a99989,
0x7a6a5a4a3a2a1a0a, 0xfaeadacabaaa9a8a,
0x7b6b5b4b3b2b1b0b, 0xfbebdbcbbbab9b8b,
0x7c6c5c4c3c2c1c0c, 0xfcecdcccbcac9c8c,
0x7d6d5d4d3d2d1d0d, 0xfdedddcdbdad9d8d
};
static const uint64_t round_const_q[] __attribute__ ((aligned (64))) =
{
0x8f9fafbfcfdfefff, 0x0f1f2f3f4f5f6f7f,
0x8e9eaebecedeeefe, 0x0e1e2e3e4e5e6e7e,
0x8d9dadbdcdddedfd, 0x0d1d2d3d4d5d6d7d,
0x8c9cacbcccdcecfc, 0x0c1c2c3c4c5c6c7c,
0x8b9babbbcbdbebfb, 0x0b1b2b3b4b5b6b7b,
0x8a9aaabacadaeafa, 0x0a1a2a3a4a5a6a7a,
0x8999a9b9c9d9e9f9, 0x0919293949596979,
0x8898a8b8c8d8e8f8, 0x0818283848586878,
0x8797a7b7c7d7e7f7, 0x0717273747576777,
0x8696a6b6c6d6e6f6, 0x0616263646566676,
0x8595a5b5c5d5e5f5, 0x0515253545556575,
0x8494a4b4c4d4e4f4, 0x0414243444546474,
0x8393a3b3c3d3e3f3, 0x0313233343536373,
0x8292a2b2c2d2e2f2, 0x0212223242526272
};
/* one round
* a0-a7 = input rows
* b0-b7 = output rows
@@ -203,22 +216,14 @@ static const uint64_t round_const_q[] __attribute__ ((aligned (64))) =
xmm8 = _mm_xor_si128( xmm8, \
casti_m128i( round_const_p, round_counter ) ); \
/* ShiftBytes P1024 + pre-AESENCLAST */\
xmm8 = _mm_shuffle_epi8( xmm8, m128_const_64( 0x0306090c0f020508, \
0x0b0e0104070a0d00 ) ); \
xmm9 = _mm_shuffle_epi8( xmm9, m128_const_64( 0x04070a0d00030609, \
0x0c0f0205080b0e01 ) ); \
xmm10 = _mm_shuffle_epi8( xmm10, m128_const_64( 0x05080b0e0104070a, \
0x0d000306090c0f02 ) ); \
xmm11 = _mm_shuffle_epi8( xmm11, m128_const_64( 0x06090c0f0205080b, \
0x0e0104070a0d0003 ) ); \
xmm12 = _mm_shuffle_epi8( xmm12, m128_const_64( 0x070a0d000306090c, \
0x0f0205080b0e0104 ) ); \
xmm13 = _mm_shuffle_epi8( xmm13, m128_const_64( 0x080b0e0104070a0d, \
0x000306090c0f0205 ) ); \
xmm14 = _mm_shuffle_epi8( xmm14, m128_const_64( 0x090c0f0205080b0e, \
0x0104070a0d000306 ) ); \
xmm15 = _mm_shuffle_epi8( xmm15, m128_const_64( 0x0e0104070a0d0003, \
0x06090c0f0205080b ) ); \
xmm8 = _mm_shuffle_epi8( xmm8, SUBSH_MASK0 ); \
xmm9 = _mm_shuffle_epi8( xmm9, SUBSH_MASK1 ); \
xmm10 = _mm_shuffle_epi8( xmm10, SUBSH_MASK2 ); \
xmm11 = _mm_shuffle_epi8( xmm11, SUBSH_MASK3 ); \
xmm12 = _mm_shuffle_epi8( xmm12, SUBSH_MASK4 ); \
xmm13 = _mm_shuffle_epi8( xmm13, SUBSH_MASK5 ); \
xmm14 = _mm_shuffle_epi8( xmm14, SUBSH_MASK6 ); \
xmm15 = _mm_shuffle_epi8( xmm15, SUBSH_MASK7 ); \
/* SubBytes + MixBytes */\
SUBMIX( xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, \
xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7 ); \
@@ -226,22 +231,14 @@ static const uint64_t round_const_q[] __attribute__ ((aligned (64))) =
/* AddRoundConstant P1024 */\
xmm0 = _mm_xor_si128( xmm0, \
casti_m128i( round_const_p, round_counter+1 ) ); \
xmm0 = _mm_shuffle_epi8( xmm0, m128_const_64( 0x0306090c0f020508, \
0x0b0e0104070a0d00 ) ); \
xmm1 = _mm_shuffle_epi8( xmm1, m128_const_64( 0x04070a0d00030609, \
0x0c0f0205080b0e01 ) ); \
xmm2 = _mm_shuffle_epi8( xmm2, m128_const_64( 0x05080b0e0104070a, \
0x0d000306090c0f02 ) ); \
xmm3 = _mm_shuffle_epi8( xmm3, m128_const_64( 0x06090c0f0205080b, \
0x0e0104070a0d0003 ) ); \
xmm4 = _mm_shuffle_epi8( xmm4, m128_const_64( 0x070a0d000306090c, \
0x0f0205080b0e0104 ) ); \
xmm5 = _mm_shuffle_epi8( xmm5, m128_const_64( 0x080b0e0104070a0d, \
0x000306090c0f0205 ) ); \
xmm6 = _mm_shuffle_epi8( xmm6, m128_const_64( 0x090c0f0205080b0e, \
0x0104070a0d000306 ) ); \
xmm7 = _mm_shuffle_epi8( xmm7, m128_const_64( 0x0e0104070a0d0003, \
0x06090c0f0205080b ) ); \
xmm0 = _mm_shuffle_epi8( xmm0, SUBSH_MASK0 ); \
xmm1 = _mm_shuffle_epi8( xmm1, SUBSH_MASK1 ); \
xmm2 = _mm_shuffle_epi8( xmm2, SUBSH_MASK2 ); \
xmm3 = _mm_shuffle_epi8( xmm3, SUBSH_MASK3 ); \
xmm4 = _mm_shuffle_epi8( xmm4, SUBSH_MASK4 ); \
xmm5 = _mm_shuffle_epi8( xmm5, SUBSH_MASK5 ); \
xmm6 = _mm_shuffle_epi8( xmm6, SUBSH_MASK6 ); \
xmm7 = _mm_shuffle_epi8( xmm7, SUBSH_MASK7 ); \
SUBMIX( xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, \
xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15 ); \
}\
@@ -262,22 +259,14 @@ static const uint64_t round_const_q[] __attribute__ ((aligned (64))) =
xmm15 = _mm_xor_si128( xmm15, \
casti_m128i( round_const_q, round_counter ) ); \
/* ShiftBytes Q1024 + pre-AESENCLAST */\
xmm8 = _mm_shuffle_epi8( xmm8, m128_const_64( 0x04070a0d00030609, \
0x0c0f0205080b0e01 ) ); \
xmm9 = _mm_shuffle_epi8( xmm9, m128_const_64( 0x06090c0f0205080b, \
0x0e0104070a0d0003 ) ); \
xmm10 = _mm_shuffle_epi8( xmm10, m128_const_64( 0x080b0e0104070a0d, \
0x000306090c0f0205 ) ); \
xmm11 = _mm_shuffle_epi8( xmm11, m128_const_64( 0x0e0104070a0d0003, \
0x06090c0f0205080b ) ); \
xmm12 = _mm_shuffle_epi8( xmm12, m128_const_64( 0x0306090c0f020508, \
0x0b0e0104070a0d00 ) ); \
xmm13 = _mm_shuffle_epi8( xmm13, m128_const_64( 0x05080b0e0104070a, \
0x0d000306090c0f02 ) ); \
xmm14 = _mm_shuffle_epi8( xmm14, m128_const_64( 0x070a0d000306090c, \
0x0f0205080b0e0104 ) ); \
xmm15 = _mm_shuffle_epi8( xmm15, m128_const_64( 0x090c0f0205080b0e, \
0x0104070a0d000306 ) ); \
xmm8 = _mm_shuffle_epi8( xmm8, SUBSH_MASK1 ); \
xmm9 = _mm_shuffle_epi8( xmm9, SUBSH_MASK3 ); \
xmm10 = _mm_shuffle_epi8( xmm10, SUBSH_MASK5 ); \
xmm11 = _mm_shuffle_epi8( xmm11, SUBSH_MASK7 ); \
xmm12 = _mm_shuffle_epi8( xmm12, SUBSH_MASK0 ); \
xmm13 = _mm_shuffle_epi8( xmm13, SUBSH_MASK2 ); \
xmm14 = _mm_shuffle_epi8( xmm14, SUBSH_MASK4 ); \
xmm15 = _mm_shuffle_epi8( xmm15, SUBSH_MASK6 ); \
/* SubBytes + MixBytes */\
SUBMIX( xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, \
xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6 , xmm7 ); \
@@ -294,22 +283,14 @@ static const uint64_t round_const_q[] __attribute__ ((aligned (64))) =
xmm7 = _mm_xor_si128( xmm7, \
casti_m128i( round_const_q, round_counter+1 ) ); \
/* ShiftBytes Q1024 + pre-AESENCLAST */\
xmm0 = _mm_shuffle_epi8( xmm0, m128_const_64( 0x04070a0d00030609, \
0x0c0f0205080b0e01 ) ); \
xmm1 = _mm_shuffle_epi8( xmm1, m128_const_64( 0x06090c0f0205080b, \
0x0e0104070a0d0003 ) ); \
xmm2 = _mm_shuffle_epi8( xmm2, m128_const_64( 0x080b0e0104070a0d, \
0x000306090c0f0205 ) ); \
xmm3 = _mm_shuffle_epi8( xmm3, m128_const_64( 0x0e0104070a0d0003, \
0x06090c0f0205080b ) ); \
xmm4 = _mm_shuffle_epi8( xmm4, m128_const_64( 0x0306090c0f020508, \
0x0b0e0104070a0d00 ) ); \
xmm5 = _mm_shuffle_epi8( xmm5, m128_const_64( 0x05080b0e0104070a, \
0x0d000306090c0f02 ) ); \
xmm6 = _mm_shuffle_epi8( xmm6, m128_const_64( 0x070a0d000306090c, \
0x0f0205080b0e0104 ) ); \
xmm7 = _mm_shuffle_epi8( xmm7, m128_const_64( 0x090c0f0205080b0e, \
0x0104070a0d000306 ) ); \
xmm0 = _mm_shuffle_epi8( xmm0, SUBSH_MASK1 ); \
xmm1 = _mm_shuffle_epi8( xmm1, SUBSH_MASK3 ); \
xmm2 = _mm_shuffle_epi8( xmm2, SUBSH_MASK5 ); \
xmm3 = _mm_shuffle_epi8( xmm3, SUBSH_MASK7 ); \
xmm4 = _mm_shuffle_epi8( xmm4, SUBSH_MASK0 ); \
xmm5 = _mm_shuffle_epi8( xmm5, SUBSH_MASK2 ); \
xmm6 = _mm_shuffle_epi8( xmm6, SUBSH_MASK4 ); \
xmm7 = _mm_shuffle_epi8( xmm7, SUBSH_MASK6 ); \
/* SubBytes + MixBytes */\
SUBMIX( xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, \
xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15 ); \
@@ -324,7 +305,7 @@ static const uint64_t round_const_q[] __attribute__ ((aligned (64))) =
* clobbers: t0-t7
*/
#define Matrix_Transpose(i0, i1, i2, i3, i4, i5, i6, i7, t0, t1, t2, t3, t4, t5, t6, t7){\
t0 = m128_const_64( 0x0f070b030e060a02, 0x0d0509010c040800 );\
t0 = TRANSP_MASK; \
\
i6 = _mm_shuffle_epi8(i6, t0);\
i0 = _mm_shuffle_epi8(i0, t0);\
@@ -412,7 +393,7 @@ static const uint64_t round_const_q[] __attribute__ ((aligned (64))) =
i4 = _mm_unpacklo_epi64(i4, i5);\
t1 = _mm_unpackhi_epi64(t1, i5);\
t2 = i6;\
o0 = m128_const_64( 0x0f070b030e060a02, 0x0d0509010c040800 ); \
o0 = TRANSP_MASK; \
i6 = _mm_unpacklo_epi64(i6, i7);\
t2 = _mm_unpackhi_epi64(t2, i7);\
/* load transpose mask into a register, because it will be used 8 times */\
@@ -653,3 +634,4 @@ void OF1024( __m128i* chaining )
return;
}
#endif

98
algo/groestl/aes_ni/groestl256-intr-aes.h
View File

@@ -11,6 +11,45 @@
#include <wmmintrin.h>
#include "hash-groestl256.h"
static const __m128i round_const_l0[] __attribute__ ((aligned (64))) =
{
{ 0x7060504030201000, 0xffffffffffffffff },
{ 0x7161514131211101, 0xffffffffffffffff },
{ 0x7262524232221202, 0xffffffffffffffff },
{ 0x7363534333231303, 0xffffffffffffffff },
{ 0x7464544434241404, 0xffffffffffffffff },
{ 0x7565554535251505, 0xffffffffffffffff },
{ 0x7666564636261606, 0xffffffffffffffff },
{ 0x7767574737271707, 0xffffffffffffffff },
{ 0x7868584838281808, 0xffffffffffffffff },
{ 0x7969594939291909, 0xffffffffffffffff }
};
static const __m128i round_const_l7[] __attribute__ ((aligned (64))) =
{
{ 0x0000000000000000, 0x8f9fafbfcfdfefff },
{ 0x0000000000000000, 0x8e9eaebecedeeefe },
{ 0x0000000000000000, 0x8d9dadbdcdddedfd },
{ 0x0000000000000000, 0x8c9cacbcccdcecfc },
{ 0x0000000000000000, 0x8b9babbbcbdbebfb },
{ 0x0000000000000000, 0x8a9aaabacadaeafa },
{ 0x0000000000000000, 0x8999a9b9c9d9e9f9 },
{ 0x0000000000000000, 0x8898a8b8c8d8e8f8 },
{ 0x0000000000000000, 0x8797a7b7c7d7e7f7 },
{ 0x0000000000000000, 0x8696a6b6c6d6e6f6 }
};
static const __m128i TRANSP_MASK = { 0x0d0509010c040800, 0x0f070b030e060a02 };
static const __m128i SUBSH_MASK0 = { 0x0c0f0104070b0e00, 0x03060a0d08020509 };
static const __m128i SUBSH_MASK1 = { 0x0e090205000d0801, 0x04070c0f0a03060b };
static const __m128i SUBSH_MASK2 = { 0x080b0306010f0a02, 0x05000e090c04070d };
static const __m128i SUBSH_MASK3 = { 0x0a0d040702090c03, 0x0601080b0e05000f };
static const __m128i SUBSH_MASK4 = { 0x0b0e0500030a0d04, 0x0702090c0f060108 };
static const __m128i SUBSH_MASK5 = { 0x0d080601040c0f05, 0x00030b0e0907020a };
static const __m128i SUBSH_MASK6 = { 0x0f0a0702050e0906, 0x01040d080b00030c };
static const __m128i SUBSH_MASK7 = { 0x090c000306080b07, 0x02050f0a0d01040e };
#define tos(a) #a
#define tostr(a) tos(a)
@@ -26,8 +65,6 @@
i = _mm_xor_si128(i, j);\
}
/**/
/* Yet another implementation of MixBytes.
This time we use the formulae (3) from the paper "Byte Slicing Groestl".
Input: a0, ..., a7
@@ -141,36 +178,6 @@
b1 = _mm_xor_si128(b1, a4);\
}/*MixBytes*/
static const uint64_t round_const_l0[] __attribute__ ((aligned (64))) =
{
0x7060504030201000, 0xffffffffffffffff,
0x7161514131211101, 0xffffffffffffffff,
0x7262524232221202, 0xffffffffffffffff,
0x7363534333231303, 0xffffffffffffffff,
0x7464544434241404, 0xffffffffffffffff,
0x7565554535251505, 0xffffffffffffffff,
0x7666564636261606, 0xffffffffffffffff,
0x7767574737271707, 0xffffffffffffffff,
0x7868584838281808, 0xffffffffffffffff,
0x7969594939291909, 0xffffffffffffffff
};
static const uint64_t round_const_l7[] __attribute__ ((aligned (64))) =
{
0x0000000000000000, 0x8f9fafbfcfdfefff,
0x0000000000000000, 0x8e9eaebecedeeefe,
0x0000000000000000, 0x8d9dadbdcdddedfd,
0x0000000000000000, 0x8c9cacbcccdcecfc,
0x0000000000000000, 0x8b9babbbcbdbebfb,
0x0000000000000000, 0x8a9aaabacadaeafa,
0x0000000000000000, 0x8999a9b9c9d9e9f9,
0x0000000000000000, 0x8898a8b8c8d8e8f8,
0x0000000000000000, 0x8797a7b7c7d7e7f7,
0x0000000000000000, 0x8696a6b6c6d6e6f6
};
/* one round
* i = round number
* a0-a7 = input rows
@@ -190,29 +197,21 @@ static const uint64_t round_const_l7[] __attribute__ ((aligned (64))) =
\
/* ShiftBytes + SubBytes (interleaved) */\
b0 = _mm_xor_si128(b0, b0);\
a0 = _mm_shuffle_epi8( a0, m128_const_64( 0x03060a0d08020509, \
0x0c0f0104070b0e00 ) ); \
a0 = _mm_shuffle_epi8( a0, SUBSH_MASK0 ); \
a0 = _mm_aesenclast_si128( a0, b0 );\
a1 = _mm_shuffle_epi8( a1, m128_const_64( 0x04070c0f0a03060b, \
0x0e090205000d0801 ) ); \
a1 = _mm_shuffle_epi8( a1, SUBSH_MASK1 ); \
a1 = _mm_aesenclast_si128( a1, b0 );\
a2 = _mm_shuffle_epi8( a2, m128_const_64( 0x05000e090c04070d, \
0x080b0306010f0a02 ) ); \
a2 = _mm_shuffle_epi8( a2, SUBSH_MASK2 ); \
a2 = _mm_aesenclast_si128( a2, b0 );\
a3 = _mm_shuffle_epi8( a3, m128_const_64( 0x0601080b0e05000f, \
0x0a0d040702090c03 ) ); \
a3 = _mm_shuffle_epi8( a3, SUBSH_MASK3 ); \
a3 = _mm_aesenclast_si128( a3, b0 );\
a4 = _mm_shuffle_epi8( a4, m128_const_64( 0x0702090c0f060108, \
0x0b0e0500030a0d04 ) ); \
a4 = _mm_shuffle_epi8( a4, SUBSH_MASK4 ); \
a4 = _mm_aesenclast_si128( a4, b0 );\
a5 = _mm_shuffle_epi8( a5, m128_const_64( 0x00030b0e0907020a, \
0x0d080601040c0f05 ) ); \
a5 = _mm_shuffle_epi8( a5, SUBSH_MASK5 ); \
a5 = _mm_aesenclast_si128( a5, b0 );\
a6 = _mm_shuffle_epi8( a6, m128_const_64( 0x01040d080b00030c, \
0x0f0a0702050e0906 ) ); \
a6 = _mm_shuffle_epi8( a6, SUBSH_MASK6 ); \
a6 = _mm_aesenclast_si128( a6, b0 );\
a7 = _mm_shuffle_epi8( a7, m128_const_64( 0x02050f0a0d01040e, \
0x090c000306080b07 ) ); \
a7 = _mm_shuffle_epi8( a7, SUBSH_MASK7 ); \
a7 = _mm_aesenclast_si128( a7, b0 );\
\
/* MixBytes */\
@@ -241,8 +240,9 @@ static const uint64_t round_const_l7[] __attribute__ ((aligned (64))) =
* outputs: i0, o1-o3
* clobbers: t0
*/
#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
t0 = m128_const_64( 0x0f070b030e060a02, 0x0d0509010c040800 ); \
t0 = TRANSP_MASK; \
\
i0 = _mm_shuffle_epi8(i0, t0);\
i1 = _mm_shuffle_epi8(i1, t0);\

92
algo/groestl/aes_ni/hash-groestl256.c
View File

@@ -214,6 +214,98 @@ HashReturn_gr update_and_final_groestl256( hashState_groestl256* ctx,
return SUCCESS_GR;
}
int groestl256_full( hashState_groestl256* ctx,
void* output, const void* input, DataLength_gr databitlen )
{
int i;
ctx->hashlen = 32;
for ( i = 0; i < SIZE256; i++ )
{
ctx->chaining[i] = _mm_setzero_si128();
ctx->buffer[i] = _mm_setzero_si128();
}
((u64*)ctx->chaining)[COLS-1] = U64BIG((u64)LENGTH);
INIT256( ctx->chaining );
ctx->buf_ptr = 0;
ctx->rem_ptr = 0;
const int len = (int)databitlen / 128;
const int hashlen_m128i = ctx->hashlen / 16; // bytes to __m128i
const int hash_offset = SIZE256 - hashlen_m128i;
int rem = ctx->rem_ptr;
int blocks = len / SIZE256;
__m128i* in = (__m128i*)input;
// --- update ---
// digest any full blocks, process directly from input
for ( i = 0; i < blocks; i++ )
TF512( ctx->chaining, &in[ i * SIZE256 ] );
ctx->buf_ptr = blocks * SIZE256;
// cryptonight has 200 byte input, an odd number of __m128i
// remainder is only 8 bytes, ie u64.
if ( databitlen % 128 !=0 )
{
// must be cryptonight, copy 64 bits of data
*(uint64_t*)(ctx->buffer) = *(uint64_t*)(&in[ ctx->buf_ptr ] );
i = -1; // signal for odd length
}
else
{
// Copy any remaining data to buffer for final transform
for ( i = 0; i < len % SIZE256; i++ )
ctx->buffer[ rem + i ] = in[ ctx->buf_ptr + i ];
i += rem; // use i as rem_ptr in final
}
//--- final ---
// adjust for final block
blocks++;
if ( i == len - 1 )
{
// all padding at once
ctx->buffer[i] = _mm_set_epi8( blocks,blocks>>8,0,0, 0,0,0,0,
0, 0,0,0, 0,0,0,0x80 );
}
else
{
if ( i == -1 )
{
// cryptonight odd length
((uint64_t*)ctx->buffer)[ 1 ] = 0x80ull;
// finish the block with zero and length padding as normal
i = 0;
}
else
{
// add first padding
ctx->buffer[i] = _mm_set_epi8( 0,0,0,0, 0,0,0,0,
0,0,0,0, 0,0,0,0x80 );
}
// add zero padding
for ( i += 1; i < SIZE256 - 1; i++ )
ctx->buffer[i] = _mm_setzero_si128();
// add length padding
// cheat since we know the block count is trivial, good if block < 256
ctx->buffer[i] = _mm_set_epi8( blocks,blocks>>8,0,0, 0,0,0,0,
0, 0,0,0, 0,0,0,0 );
}
// digest final padding block and do output transform
TF512( ctx->chaining, ctx->buffer );
OF512( ctx->chaining );
// store hash result in output
for ( i = 0; i < hashlen_m128i; i++ )
casti_m128i( output, i ) = ctx->chaining[ hash_offset + i ];
return SUCCESS_GR;
}
/* hash bit sequence */
HashReturn_gr hash_groestl256(int hashbitlen,
const BitSequence_gr* data,

3
algo/groestl/aes_ni/hash-groestl256.h
View File

@@ -115,4 +115,7 @@ HashReturn_gr hash_groestli256( int, const BitSequence_gr*, DataLength_gr,
HashReturn_gr update_and_final_groestl256( hashState_groestl256*, void*,
const void*, DataLength_gr );
int groestl256_full( hashState_groestl256* ctx,
void* output, const void* input, DataLength_gr databitlen );
#endif /* __hash_h */

5
algo/groestl/groestl.c
View File

@@ -1,4 +1,7 @@
#include "groestl-gate.h"
#if !defined(GROESTL_8WAY) && !defined(GROESTLX16R_4WAY)
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
@@ -88,4 +91,4 @@ int scanhash_groestl( struct work *work, uint32_t max_nonce,
*hashes_done = pdata[19] - first_nonce + 1;
return 0;
}
#endif

79
algo/groestl/groestl256-hash-4way.c
View File

@@ -23,7 +23,6 @@ int groestl256_4way_init( groestl256_4way_context* ctx, uint64_t hashlen )
int i;
ctx->hashlen = hashlen;
SET_CONSTANTS();
if (ctx->chaining == NULL || ctx->buffer == NULL)
return 1;
@@ -36,9 +35,6 @@ int groestl256_4way_init( groestl256_4way_context* ctx, uint64_t hashlen )
// The only non-zero in the IV is len. It can be hard coded.
ctx->chaining[ 3 ] = m512_const2_64( 0, 0x0100000000000000 );
// uint64_t len = U64BIG((uint64_t)LENGTH);
// ctx->chaining[ COLS/2 -1 ] = _mm512_set4_epi64( len, 0, len, 0 );
// INIT256_4way(ctx->chaining);
ctx->buf_ptr = 0;
ctx->rem_ptr = 0;
@@ -46,6 +42,77 @@ int groestl256_4way_init( groestl256_4way_context* ctx, uint64_t hashlen )
return 0;
}
int groestl256_4way_full( groestl256_4way_context* ctx, void* output,
const void* input, uint64_t databitlen )
{
const int len = (int)databitlen / 128;
const int hashlen_m128i = 32 / 16; // bytes to __m128i
const int hash_offset = SIZE256 - hashlen_m128i;
int rem = ctx->rem_ptr;
int blocks = len / SIZE256;
__m512i* in = (__m512i*)input;
int i;
if (ctx->chaining == NULL || ctx->buffer == NULL)
return 1;
for ( i = 0; i < SIZE256; i++ )
{
ctx->chaining[i] = m512_zero;
ctx->buffer[i] = m512_zero;
}
// The only non-zero in the IV is len. It can be hard coded.
ctx->chaining[ 3 ] = m512_const2_64( 0, 0x0100000000000000 );
ctx->buf_ptr = 0;
ctx->rem_ptr = 0;
// --- update ---
// digest any full blocks, process directly from input
for ( i = 0; i < blocks; i++ )
TF512_4way( ctx->chaining, &in[ i * SIZE256 ] );
ctx->buf_ptr = blocks * SIZE256;
// copy any remaining data to buffer, it may already contain data
// from a previous update for a midstate precalc
for ( i = 0; i < len % SIZE256; i++ )
ctx->buffer[ rem + i ] = in[ ctx->buf_ptr + i ];
i += rem; // use i as rem_ptr in final
//--- final ---
blocks++; // adjust for final block
if ( i == SIZE256 - 1 )
{
// only 1 vector left in buffer, all padding at once
ctx->buffer[i] = m512_const2_64( (uint64_t)blocks << 56, 0x80 );
}
else
{
// add first padding
ctx->buffer[i] = m512_const4_64( 0, 0x80, 0, 0x80 );
// add zero padding
for ( i += 1; i < SIZE256 - 1; i++ )
ctx->buffer[i] = m512_zero;
// add length padding, second last byte is zero unless blocks > 255
ctx->buffer[i] = m512_const2_64( (uint64_t)blocks << 56, 0 );
}
// digest final padding block and do output transform
TF512_4way( ctx->chaining, ctx->buffer );
OF512_4way( ctx->chaining );
// store hash result in output
for ( i = 0; i < hashlen_m128i; i++ )
casti_m512i( output, i ) = ctx->chaining[ hash_offset + i ];
return 0;
}
int groestl256_4way_update_close( groestl256_4way_context* ctx, void* output,
const void* input, uint64_t databitlen )
{
@@ -75,11 +142,11 @@ int groestl256_4way_update_close( groestl256_4way_context* ctx, void* output,
blocks++; // adjust for final block
if ( i == SIZE256 - 1 )
{
{
// only 1 vector left in buffer, all padding at once
ctx->buffer[i] = m512_const1_128( _mm_set_epi8(
blocks, blocks>>8,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0x80 ) );
}
}
else
{
// add first padding

3
algo/groestl/groestl256-hash-4way.h
View File

@@ -71,5 +71,8 @@ int groestl256_4way_init( groestl256_4way_context*, uint64_t );
int groestl256_4way_update_close( groestl256_4way_context*, void*,
const void*, uint64_t );
int groestl256_4way_full( groestl256_4way_context*, void*,
const void*, uint64_t );
#endif
#endif

187
algo/groestl/groestl256-intr-4way.h
View File

@@ -14,17 +14,78 @@
#include "groestl256-hash-4way.h"
#if defined(__VAES__)
static const __m128i round_const_l0[] __attribute__ ((aligned (64))) =
{
{ 0x7060504030201000, 0xffffffffffffffff },
{ 0x7161514131211101, 0xffffffffffffffff },
{ 0x7262524232221202, 0xffffffffffffffff },
{ 0x7363534333231303, 0xffffffffffffffff },
{ 0x7464544434241404, 0xffffffffffffffff },
{ 0x7565554535251505, 0xffffffffffffffff },
{ 0x7666564636261606, 0xffffffffffffffff },
{ 0x7767574737271707, 0xffffffffffffffff },
{ 0x7868584838281808, 0xffffffffffffffff },
{ 0x7969594939291909, 0xffffffffffffffff }
};
/* global constants */
__m512i ROUND_CONST_Lx;
__m512i ROUND_CONST_L0[ROUNDS512];
__m512i ROUND_CONST_L7[ROUNDS512];
//__m512i ROUND_CONST_P[ROUNDS1024];
//__m512i ROUND_CONST_Q[ROUNDS1024];
__m512i TRANSP_MASK;
__m512i SUBSH_MASK[8];
__m512i ALL_1B;
__m512i ALL_FF;
static const __m128i round_const_l7[] __attribute__ ((aligned (64))) =
{
{ 0x0000000000000000, 0x8f9fafbfcfdfefff },
{ 0x0000000000000000, 0x8e9eaebecedeeefe },
{ 0x0000000000000000, 0x8d9dadbdcdddedfd },
{ 0x0000000000000000, 0x8c9cacbcccdcecfc },
{ 0x0000000000000000, 0x8b9babbbcbdbebfb },
{ 0x0000000000000000, 0x8a9aaabacadaeafa },
{ 0x0000000000000000, 0x8999a9b9c9d9e9f9 },
{ 0x0000000000000000, 0x8898a8b8c8d8e8f8 },
{ 0x0000000000000000, 0x8797a7b7c7d7e7f7 },
{ 0x0000000000000000, 0x8696a6b6c6d6e6f6 }
};
static const __m512i TRANSP_MASK = { 0x0d0509010c040800, 0x0f070b030e060a02,
0x1d1519111c141810, 0x1f171b131e161a12,
0x2d2529212c242820, 0x2f272b232e262a22,
0x3d3539313c343830, 0x3f373b333e363a32 };
static const __m512i SUBSH_MASK0 = { 0x0c0f0104070b0e00, 0x03060a0d08020509,
0x1c1f1114171b1e10, 0x13161a1d18121519,
0x2c2f2124272b2e20, 0x23262a2d28222529,
0x3c3f3134373b3e30, 0x33363a3d38323539 };
static const __m512i SUBSH_MASK1 = { 0x0e090205000d0801, 0x04070c0f0a03060b,
0x1e191215101d1801, 0x14171c1f1a13161b,
0x2e292225202d2821, 0x24272c2f2a23262b,
0x3e393235303d3831, 0x34373c3f3a33363b };
static const __m512i SUBSH_MASK2 = { 0x080b0306010f0a02, 0x05000e090c04070d,
0x181b1316111f1a12, 0x15101e191c14171d,
0x282b2326212f2a22, 0x25202e292c24272d,
0x383b3336313f3a32, 0x35303e393c34373d };
static const __m512i SUBSH_MASK3 = { 0x0a0d040702090c03, 0x0601080b0e05000f,
0x1a1d141712191c13, 0x1611181b1e15101f,
0x2a2d242722292c23, 0x2621282b2e25202f,
0x3a3d343732393c33, 0x3631383b3e35303f };
static const __m512i SUBSH_MASK4 = { 0x0b0e0500030a0d04, 0x0702090c0f060108,
0x1b1e1510131a1d14, 0x1712191c1f161118,
0x2b2e2520232a2d24, 0x2722292c2f262128,
0x3b3e3530333a3d34, 0x3732393c3f363138 };
static const __m512i SUBSH_MASK5 = { 0x0d080601040c0f05, 0x00030b0e0907020a,
0x1d181611141c1f15, 0x10131b1e1917121a,
0x2d282621242c2f25, 0x20232b2e2927222a,
0x3d383631343c3f35, 0x30333b3e3937323a };
static const __m512i SUBSH_MASK6 = { 0x0f0a0702050e0906, 0x01040d080b00030c,
0x1f1a1712151e1916, 0x11141d181b10131c,
0x2f2a2722252e2926, 0x21242d282b20232c,
0x3f3a3732353e3936, 0x31343d383b30333c };
static const __m512i SUBSH_MASK7 = { 0x090c000306080b07, 0x02050f0a0d01040e,
0x191c101316181b17, 0x12151f1a1d11141e,
0x292c202326282b27, 0x22252f2a2d21242e,
0x393c303336383b37, 0x32353f3a3d31343e };
#define tos(a) #a
#define tostr(a) tos(a)
@@ -40,8 +101,6 @@ __m512i ALL_FF;
i = _mm512_xor_si512(i, j);\
}
/**/
/* Yet another implementation of MixBytes.
This time we use the formulae (3) from the paper "Byte Slicing Groestl".
Input: a0, ..., a7
@@ -155,95 +214,36 @@ __m512i ALL_FF;
b1 = _mm512_xor_si512(b1, a4);\
}/*MixBytes*/
// calculate the round constants seperately and load at startup
#define SET_CONSTANTS(){\
ALL_1B = _mm512_set1_epi32( 0x1b1b1b1b );\
TRANSP_MASK = _mm512_set_epi32( \
0x3f373b33, 0x3e363a32, 0x3d353931, 0x3c343830, \
0x2f272b23, 0x2e262a22, 0x2d252921, 0x2c242820, \
0x1f171b13, 0x1e161a12, 0x1d151911, 0x1c141810, \
0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800 ); \
SUBSH_MASK[0] = _mm512_set_epi32( \
0x33363a3d, 0x38323539, 0x3c3f3134, 0x373b3e30, \
0x23262a2d, 0x28222529, 0x2c2f2124, 0x272b2e20, \
0x13161a1d, 0x18121519, 0x1c1f1114, 0x171b1e10, \
0x03060a0d, 0x08020509, 0x0c0f0104, 0x070b0e00 ); \
SUBSH_MASK[1] = _mm512_set_epi32( \
0x34373c3f, 0x3a33363b, 0x3e393235, 0x303d3831, \
0x24272c2f, 0x2a23262b, 0x2e292225, 0x202d2821, \
0x14171c1f, 0x1a13161b, 0x1e191215, 0x101d1801, \
0x04070c0f, 0x0a03060b, 0x0e090205, 0x000d0801 );\
SUBSH_MASK[2] = _mm512_set_epi32( \
0x35303e39, 0x3c34373d, 0x383b3336, 0x313f3a32, \
0x25202e29, 0x2c24272d, 0x282b2326, 0x212f2a22, \
0x15101e19, 0x1c14171d, 0x181b1316, 0x111f1a12, \
0x05000e09, 0x0c04070d, 0x080b0306, 0x010f0a02 );\
SUBSH_MASK[3] = _mm512_set_epi32( \
0x3631383b, 0x3e35303f, 0x3a3d3437, 0x32393c33, \
0x2621282b, 0x2e25202f, 0x2a2d2427, 0x22292c23, \
0x1611181b, 0x1e15101f, 0x1a1d1417, 0x12191c13, \
0x0601080b, 0x0e05000f, 0x0a0d0407, 0x02090c03 );\
SUBSH_MASK[4] = _mm512_set_epi32( \
0x3732393c, 0x3f363138, 0x3b3e3530, 0x333a3d34, \
0x2722292c, 0x2f262128, 0x2b2e2520, 0x232a2d24, \
0x1712191c, 0x1f161118, 0x1b1e1510, 0x131a1d14, \
0x0702090c, 0x0f060108, 0x0b0e0500, 0x030a0d04 );\
SUBSH_MASK[5] = _mm512_set_epi32( \
0x30333b3e, 0x3937323a, 0x3d383631, 0x343c3f35, \
0x20232b2e, 0x2927222a, 0x2d282621, 0x242c2f25, \
0x10131b1e, 0x1917121a, 0x1d181611, 0x141c1f15, \
0x00030b0e, 0x0907020a, 0x0d080601, 0x040c0f05 );\
SUBSH_MASK[6] = _mm512_set_epi32( \
0x31343d38, 0x3b30333c, 0x3f3a3732, 0x353e3936, \
0x21242d28, 0x2b20232c, 0x2f2a2722, 0x252e2926, \
0x11141d18, 0x1b10131c, 0x1f1a1712, 0x151e1916, \
0x01040d08, 0x0b00030c, 0x0f0a0702, 0x050e0906 );\
SUBSH_MASK[7] = _mm512_set_epi32( \
0x32353f3a, 0x3d31343e, 0x393c3033, 0x36383b37, \
0x22252f2a, 0x2d21242e, 0x292c2023, 0x26282b27, \
0x12151f1a, 0x1d11141e, 0x191c1013, 0x16181b17, \
0x02050f0a, 0x0d01040e, 0x090c0003, 0x06080b07 );\
for ( i = 0; i < ROUNDS512; i++ ) \
{\
ROUND_CONST_L0[i] = _mm512_set4_epi32( 0xffffffff, 0xffffffff, \
0x70605040 ^ ( i * 0x01010101 ), 0x30201000 ^ ( i * 0x01010101 ) ); \
ROUND_CONST_L7[i] = _mm512_set4_epi32( 0x8f9fafbf ^ ( i * 0x01010101 ), \
0xcfdfefff ^ ( i * 0x01010101 ), 0x00000000, 0x00000000 ); \
}\
ROUND_CONST_Lx = _mm512_set4_epi32( 0xffffffff, 0xffffffff, \
0x00000000, 0x00000000 ); \
}while(0);\
#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
/* AddRoundConstant */\
b1 = ROUND_CONST_Lx;\
a0 = _mm512_xor_si512( a0, (ROUND_CONST_L0[i]) );\
b1 = m512_const2_64( 0xffffffffffffffff, 0 ); \
a0 = _mm512_xor_si512( a0, m512_const1_128( round_const_l0[i] ) );\
a1 = _mm512_xor_si512( a1, b1 );\
a2 = _mm512_xor_si512( a2, b1 );\
a3 = _mm512_xor_si512( a3, b1 );\
a4 = _mm512_xor_si512( a4, b1 );\
a5 = _mm512_xor_si512( a5, b1 );\
a6 = _mm512_xor_si512( a6, b1 );\
a7 = _mm512_xor_si512( a7, (ROUND_CONST_L7[i]) );\
a7 = _mm512_xor_si512( a7, m512_const1_128( round_const_l7[i] ) );\
\
/* ShiftBytes + SubBytes (interleaved) */\
b0 = _mm512_xor_si512( b0, b0 );\
a0 = _mm512_shuffle_epi8( a0, (SUBSH_MASK[0]) );\
a0 = _mm512_shuffle_epi8( a0, SUBSH_MASK0 );\
a0 = _mm512_aesenclast_epi128(a0, b0 );\
a1 = _mm512_shuffle_epi8( a1, (SUBSH_MASK[1]) );\
a1 = _mm512_shuffle_epi8( a1, SUBSH_MASK1 );\
a1 = _mm512_aesenclast_epi128(a1, b0 );\
a2 = _mm512_shuffle_epi8( a2, (SUBSH_MASK[2]) );\
a2 = _mm512_shuffle_epi8( a2, SUBSH_MASK2 );\
a2 = _mm512_aesenclast_epi128(a2, b0 );\
a3 = _mm512_shuffle_epi8( a3, (SUBSH_MASK[3]) );\
a3 = _mm512_shuffle_epi8( a3, SUBSH_MASK3 );\
a3 = _mm512_aesenclast_epi128(a3, b0 );\
a4 = _mm512_shuffle_epi8( a4, (SUBSH_MASK[4]) );\
a4 = _mm512_shuffle_epi8( a4, SUBSH_MASK4 );\
a4 = _mm512_aesenclast_epi128(a4, b0 );\
a5 = _mm512_shuffle_epi8( a5, (SUBSH_MASK[5]) );\
a5 = _mm512_shuffle_epi8( a5, SUBSH_MASK5 );\
a5 = _mm512_aesenclast_epi128(a5, b0 );\
a6 = _mm512_shuffle_epi8( a6, (SUBSH_MASK[6]) );\
a6 = _mm512_shuffle_epi8( a6, SUBSH_MASK6 );\
a6 = _mm512_aesenclast_epi128(a6, b0 );\
a7 = _mm512_shuffle_epi8( a7, (SUBSH_MASK[7]) );\
a7 = _mm512_shuffle_epi8( a7, SUBSH_MASK7 );\
a7 = _mm512_aesenclast_epi128( a7, b0 );\
\
/* MixBytes */\
@@ -390,29 +390,6 @@ __m512i ALL_FF;
}/**/
void INIT256_4way( __m512i* chaining )
{
static __m512i xmm0, xmm2, xmm6, xmm7;
static __m512i xmm12, xmm13, xmm14, xmm15;
/* load IV into registers xmm12 - xmm15 */
xmm12 = chaining[0];
xmm13 = chaining[1];
xmm14 = chaining[2];
xmm15 = chaining[3];
/* transform chaining value from column ordering into row ordering */
/* we put two rows (64 bit) of the IV into one 128-bit XMM register */
Matrix_Transpose_A(xmm12, xmm13, xmm14, xmm15, xmm2, xmm6, xmm7, xmm0);
/* store transposed IV */
chaining[0] = xmm12;
chaining[1] = xmm2;
chaining[2] = xmm6;
chaining[3] = xmm7;
}
void TF512_4way( __m512i* chaining, __m512i* message )
{
static __m512i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;

5
algo/groestl/groestl512-hash-4way.c
View File

@@ -19,10 +19,6 @@
int groestl512_4way_init( groestl512_4way_context* ctx, uint64_t hashlen )
{
int i;
SET_CONSTANTS();
if (ctx->chaining == NULL || ctx->buffer == NULL)
return 1;
@@ -99,7 +95,6 @@ int groestl512_4way_full( groestl512_4way_context* ctx, void* output,
// --- init ---
SET_CONSTANTS();
memset_zero_512( ctx->chaining, SIZE512 );
memset_zero_512( ctx->buffer, SIZE512 );
ctx->chaining[ 6 ] = m512_const2_64( 0x0200000000000000, 0 );

229
algo/groestl/groestl512-intr-4way.h
View File

@@ -15,16 +15,86 @@
#if defined(__VAES__)
/* global constants */
__m512i ROUND_CONST_Lx;
//__m128i ROUND_CONST_L0[ROUNDS512];
//__m128i ROUND_CONST_L7[ROUNDS512];
__m512i ROUND_CONST_P[ROUNDS1024];
__m512i ROUND_CONST_Q[ROUNDS1024];
__m512i TRANSP_MASK;
__m512i SUBSH_MASK[8];
__m512i ALL_1B;
__m512i ALL_FF;
static const __m128i round_const_p[] __attribute__ ((aligned (64))) =
{
{ 0x7060504030201000, 0xf0e0d0c0b0a09080 },
{ 0x7161514131211101, 0xf1e1d1c1b1a19181 },
{ 0x7262524232221202, 0xf2e2d2c2b2a29282 },
{ 0x7363534333231303, 0xf3e3d3c3b3a39383 },
{ 0x7464544434241404, 0xf4e4d4c4b4a49484 },
{ 0x7565554535251505, 0xf5e5d5c5b5a59585 },
{ 0x7666564636261606, 0xf6e6d6c6b6a69686 },
{ 0x7767574737271707, 0xf7e7d7c7b7a79787 },
{ 0x7868584838281808, 0xf8e8d8c8b8a89888 },
{ 0x7969594939291909, 0xf9e9d9c9b9a99989 },
{ 0x7a6a5a4a3a2a1a0a, 0xfaeadacabaaa9a8a },
{ 0x7b6b5b4b3b2b1b0b, 0xfbebdbcbbbab9b8b },
{ 0x7c6c5c4c3c2c1c0c, 0xfcecdcccbcac9c8c },
{ 0x7d6d5d4d3d2d1d0d, 0xfdedddcdbdad9d8d }
};
static const __m128i round_const_q[] __attribute__ ((aligned (64))) =
{
{ 0x8f9fafbfcfdfefff, 0x0f1f2f3f4f5f6f7f },
{ 0x8e9eaebecedeeefe, 0x0e1e2e3e4e5e6e7e },
{ 0x8d9dadbdcdddedfd, 0x0d1d2d3d4d5d6d7d },
{ 0x8c9cacbcccdcecfc, 0x0c1c2c3c4c5c6c7c },
{ 0x8b9babbbcbdbebfb, 0x0b1b2b3b4b5b6b7b },
{ 0x8a9aaabacadaeafa, 0x0a1a2a3a4a5a6a7a },
{ 0x8999a9b9c9d9e9f9, 0x0919293949596979 },
{ 0x8898a8b8c8d8e8f8, 0x0818283848586878 },
{ 0x8797a7b7c7d7e7f7, 0x0717273747576777 },
{ 0x8696a6b6c6d6e6f6, 0x0616263646566676 },
{ 0x8595a5b5c5d5e5f5, 0x0515253545556575 },
{ 0x8494a4b4c4d4e4f4, 0x0414243444546474 },
{ 0x8393a3b3c3d3e3f3, 0x0313233343536373 },
{ 0x8292a2b2c2d2e2f2, 0x0212223242526272 }
};
static const __m512i TRANSP_MASK = { 0x0d0509010c040800, 0x0f070b030e060a02,
0x1d1519111c141810, 0x1f171b131e161a12,
0x2d2529212c242820, 0x2f272b232e262a22,
0x3d3539313c343830, 0x3f373b333e363a32 };
static const __m512i SUBSH_MASK0 = { 0x0b0e0104070a0d00, 0x0306090c0f020508,
0x1b1e1114171a1d10, 0x1316191c1f121518,
0x2b2e2124272a2d20, 0x2326292c2f222528,
0x3b3e3134373a3d30, 0x3336393c3f323538 };
static const __m512i SUBSH_MASK1 = { 0x0c0f0205080b0e01, 0x04070a0d00030609,
0x1c1f1215181b1e11, 0x14171a1d10131619,
0x2c2f2225282b2e21, 0x24272a2d20232629,
0x3c3f3235383b3e31, 0x34373a3d30333639 };
static const __m512i SUBSH_MASK2 = { 0x0d000306090c0f02, 0x05080b0e0104070a,
0x1d101316191c1f12, 0x15181b1e1114171a,
0x2d202326292c2f22, 0x25282b2e2124272a,
0x3d303336393c3f32, 0x35383b3e3134373a };
static const __m512i SUBSH_MASK3 = { 0x0e0104070a0d0003, 0x06090c0f0205080b,
0x1e1114171a1d1013, 0x16191c1f1215181b,
0x2e2124272a2d2023, 0x26292c2f2225282b,
0x3e3134373a3d3033, 0x36393c3f3235383b };
static const __m512i SUBSH_MASK4 = { 0x0f0205080b0e0104, 0x070a0d000306090c,
0x1f1215181b1e1114, 0x171a1d101316191c,
0x2f2225282b2e2124, 0x272a2d202326292c,
0x3f3235383b3e3134, 0x373a3d303336393c };
static const __m512i SUBSH_MASK5 = { 0x000306090c0f0205, 0x080b0e0104070a0d,
0x101316191c1f1215, 0x181b1e1114171a1d,
0x202326292c2f2225, 0x282b2e2124272a2d,
0x303336393c3f3235, 0x383b3e3134373a3d };
static const __m512i SUBSH_MASK6 = { 0x0104070a0d000306, 0x090c0f0205080b0e,
0x1114171a1d101316, 0x191c1f1215181b1e,
0x2124272a2d202326, 0x292c2f2225282b2e,
0x3134373a3d303336, 0x393c3f3235383b3e };
static const __m512i SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003,
0x16191c1f1215181b, 0x1e1114171a1d1013,
0x26292c2f2225282b, 0x2e2124272a2d2023,
0x36393c3f3235383b, 0x3e3134373a3d3033 };
#define tos(a) #a
#define tostr(a) tos(a)
@@ -155,69 +225,6 @@ __m512i ALL_FF;
b1 = _mm512_xor_si512(b1, a4);\
}/*MixBytes*/
// calculate the round constants seperately and load at startup
#define SET_CONSTANTS(){\
ALL_FF = _mm512_set1_epi32( 0xffffffff );\
ALL_1B = _mm512_set1_epi32( 0x1b1b1b1b );\
TRANSP_MASK = _mm512_set_epi32( \
0x3f373b33, 0x3e363a32, 0x3d353931, 0x3c343830, \
0x2f272b23, 0x2e262a22, 0x2d252921, 0x2c242820, \
0x1f171b13, 0x1e161a12, 0x1d151911, 0x1c141810, \
0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800 ); \
SUBSH_MASK[0] = _mm512_set_epi32( \
0x3336393c, 0x3f323538, 0x3b3e3134, 0x373a3d30, \
0x2326292c, 0x2f222528, 0x2b2e2124, 0x272a2d20, \
0x1316191c, 0x1f121518, 0x1b1e1114, 0x171a1d10, \
0x0306090c, 0x0f020508, 0x0b0e0104, 0x070a0d00 ); \
SUBSH_MASK[1] = _mm512_set_epi32( \
0x34373a3d, 0x30333639, 0x3c3f3235, 0x383b3e31, \
0x24272a2d, 0x20232629, 0x2c2f2225, 0x282b2e21, \
0x14171a1d, 0x10131619, 0x1c1f1215, 0x181b1e11, \
0x04070a0d, 0x00030609, 0x0c0f0205, 0x080b0e01 ); \
SUBSH_MASK[2] = _mm512_set_epi32( \
0x35383b3e, 0x3134373a, 0x3d303336, 0x393c3f32, \
0x25282b2e, 0x2124272a, 0x2d202326, 0x292c2f22, \
0x15181b1e, 0x1114171a, 0x1d101316, 0x191c1f12, \
0x05080b0e, 0x0104070a, 0x0d000306, 0x090c0f02 ); \
SUBSH_MASK[3] = _mm512_set_epi32( \
0x36393c3f, 0x3235383b, 0x3e313437, 0x3a3d3033, \
0x26292c2f, 0x2225282b, 0x2e212427, 0x2a2d2023, \
0x16191c1f, 0x1215181b, 0x1e111417, 0x1a1d1013, \
0x06090c0f, 0x0205080b, 0x0e010407, 0x0a0d0003 ); \
SUBSH_MASK[4] = _mm512_set_epi32( \
0x373a3d30, 0x3336393c, 0x3f323538, 0x3b3e3134, \
0x272a2d20, 0x2326292c, 0x2f222528, 0x2b2e2124, \
0x171a1d10, 0x1316191c, 0x1f121518, 0x1b1e1114, \
0x070a0d00, 0x0306090c, 0x0f020508, 0x0b0e0104 ); \
SUBSH_MASK[5] = _mm512_set_epi32( \
0x383b3e31, 0x34373a3d, 0x30333639, 0x3c3f3235, \
0x282b2e21, 0x24272a2d, 0x20232629, 0x2c2f2225, \
0x181b1e11, 0x14171a1d, 0x10131619, 0x1c1f1215, \
0x080b0e01, 0x04070a0d, 0x00030609, 0x0c0f0205 ); \
SUBSH_MASK[6] = _mm512_set_epi32( \
0x393c3f32, 0x35383b3e, 0x3134373a, 0x3d303336, \
0x292c2f22, 0x25282b2e, 0x2124272a, 0x2d202326, \
0x191c1f12, 0x15181b1e, 0x1114171a, 0x1d101316, \
0x090c0f02, 0x05080b0e, 0x0104070a, 0x0d000306 ); \
SUBSH_MASK[7] = _mm512_set_epi32( \
0x3e313437, 0x3a3d3033, 0x36393c3f, 0x3235383b, \
0x2e212427, 0x2a2d2023, 0x26292c2f, 0x2225282b, \
0x1e111417, 0x1a1d1013, 0x16191c1f, 0x1215181b, \
0x0e010407, 0x0a0d0003, 0x06090c0f, 0x0205080b ); \
for( i = 0; i < ROUNDS1024; i++ ) \
{ \
ROUND_CONST_P[i] = _mm512_set4_epi32( 0xf0e0d0c0 ^ (i * 0x01010101), \
0xb0a09080 ^ (i * 0x01010101), \
0x70605040 ^ (i * 0x01010101), \
0x30201000 ^ (i * 0x01010101) ); \
ROUND_CONST_Q[i] = _mm512_set4_epi32( 0x0f1f2f3f ^ (i * 0x01010101), \
0x4f5f6f7f ^ (i * 0x01010101), \
0x8f9fafbf ^ (i * 0x01010101), \
0xcfdfefff ^ (i * 0x01010101));\
} \
}while(0);\
/* one round
* a0-a7 = input rows
* b0-b7 = output rows
@@ -242,30 +249,32 @@ __m512i ALL_FF;
for ( round_counter = 0; round_counter < 14; round_counter += 2 ) \
{ \
/* AddRoundConstant P1024 */\
xmm8 = _mm512_xor_si512( xmm8, ( ROUND_CONST_P[ round_counter ] ) );\
xmm8 = _mm512_xor_si512( xmm8, m512_const1_128( \
casti_m128i( round_const_p, round_counter ) ) ); \
/* ShiftBytes P1024 + pre-AESENCLAST */\
xmm8 = _mm512_shuffle_epi8( xmm8, ( SUBSH_MASK[0] ) );\
xmm9 = _mm512_shuffle_epi8( xmm9, ( SUBSH_MASK[1] ) );\
xmm10 = _mm512_shuffle_epi8( xmm10, ( SUBSH_MASK[2] ) );\
xmm11 = _mm512_shuffle_epi8( xmm11, ( SUBSH_MASK[3] ) );\
xmm12 = _mm512_shuffle_epi8( xmm12, ( SUBSH_MASK[4] ) );\
xmm13 = _mm512_shuffle_epi8( xmm13, ( SUBSH_MASK[5] ) );\
xmm14 = _mm512_shuffle_epi8( xmm14, ( SUBSH_MASK[6] ) );\
xmm15 = _mm512_shuffle_epi8( xmm15, ( SUBSH_MASK[7] ) );\
xmm8 = _mm512_shuffle_epi8( xmm8, SUBSH_MASK0 ); \
xmm9 = _mm512_shuffle_epi8( xmm9, SUBSH_MASK1 );\
xmm10 = _mm512_shuffle_epi8( xmm10, SUBSH_MASK2 );\
xmm11 = _mm512_shuffle_epi8( xmm11, SUBSH_MASK3 );\
xmm12 = _mm512_shuffle_epi8( xmm12, SUBSH_MASK4 );\
xmm13 = _mm512_shuffle_epi8( xmm13, SUBSH_MASK5 );\
xmm14 = _mm512_shuffle_epi8( xmm14, SUBSH_MASK6 );\
xmm15 = _mm512_shuffle_epi8( xmm15, SUBSH_MASK7 );\
/* SubBytes + MixBytes */\
SUBMIX(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
\
/* AddRoundConstant P1024 */\
xmm0 = _mm512_xor_si512( xmm0, ( ROUND_CONST_P[ round_counter+1 ] ) );\
xmm0 = _mm512_xor_si512( xmm0, m512_const1_128( \
casti_m128i( round_const_p, round_counter+1 ) ) ); \
/* ShiftBytes P1024 + pre-AESENCLAST */\
xmm0 = _mm512_shuffle_epi8( xmm0, ( SUBSH_MASK[0] ) );\
xmm1 = _mm512_shuffle_epi8( xmm1, ( SUBSH_MASK[1] ) );\
xmm2 = _mm512_shuffle_epi8( xmm2, ( SUBSH_MASK[2] ) );\
xmm3 = _mm512_shuffle_epi8( xmm3, ( SUBSH_MASK[3] ) );\
xmm4 = _mm512_shuffle_epi8( xmm4, ( SUBSH_MASK[4] ) );\
xmm5 = _mm512_shuffle_epi8( xmm5, ( SUBSH_MASK[5] ) );\
xmm6 = _mm512_shuffle_epi8( xmm6, ( SUBSH_MASK[6] ) );\
xmm7 = _mm512_shuffle_epi8( xmm7, ( SUBSH_MASK[7] ) );\
xmm0 = _mm512_shuffle_epi8( xmm0, SUBSH_MASK0 );\
xmm1 = _mm512_shuffle_epi8( xmm1, SUBSH_MASK1 );\
xmm2 = _mm512_shuffle_epi8( xmm2, SUBSH_MASK2 );\
xmm3 = _mm512_shuffle_epi8( xmm3, SUBSH_MASK3 );\
xmm4 = _mm512_shuffle_epi8( xmm4, SUBSH_MASK4 );\
xmm5 = _mm512_shuffle_epi8( xmm5, SUBSH_MASK5 );\
xmm6 = _mm512_shuffle_epi8( xmm6, SUBSH_MASK6 );\
xmm7 = _mm512_shuffle_epi8( xmm7, SUBSH_MASK7 );\
/* SubBytes + MixBytes */\
SUBMIX(xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
}\
@@ -284,16 +293,17 @@ __m512i ALL_FF;
xmm12 = _mm512_xor_si512( xmm12, xmm1 );\
xmm13 = _mm512_xor_si512( xmm13, xmm1 );\
xmm14 = _mm512_xor_si512( xmm14, xmm1 );\
xmm15 = _mm512_xor_si512( xmm15, ( ROUND_CONST_Q[ round_counter ] ) );\
xmm15 = _mm512_xor_si512( xmm15, m512_const1_128( \
casti_m128i( round_const_q, round_counter ) ) ); \
/* ShiftBytes Q1024 + pre-AESENCLAST */\
xmm8 = _mm512_shuffle_epi8( xmm8, ( SUBSH_MASK[1] ) );\
xmm9 = _mm512_shuffle_epi8( xmm9, ( SUBSH_MASK[3] ) );\
xmm10 = _mm512_shuffle_epi8( xmm10, ( SUBSH_MASK[5] ) );\
xmm11 = _mm512_shuffle_epi8( xmm11, ( SUBSH_MASK[7] ) );\
xmm12 = _mm512_shuffle_epi8( xmm12, ( SUBSH_MASK[0] ) );\
xmm13 = _mm512_shuffle_epi8( xmm13, ( SUBSH_MASK[2] ) );\
xmm14 = _mm512_shuffle_epi8( xmm14, ( SUBSH_MASK[4] ) );\
xmm15 = _mm512_shuffle_epi8( xmm15, ( SUBSH_MASK[6] ) );\
xmm8 = _mm512_shuffle_epi8( xmm8, SUBSH_MASK1 );\
xmm9 = _mm512_shuffle_epi8( xmm9, SUBSH_MASK3 );\
xmm10 = _mm512_shuffle_epi8( xmm10, SUBSH_MASK5 );\
xmm11 = _mm512_shuffle_epi8( xmm11, SUBSH_MASK7 );\
xmm12 = _mm512_shuffle_epi8( xmm12, SUBSH_MASK0 );\
xmm13 = _mm512_shuffle_epi8( xmm13, SUBSH_MASK2 );\
xmm14 = _mm512_shuffle_epi8( xmm14, SUBSH_MASK4 );\
xmm15 = _mm512_shuffle_epi8( xmm15, SUBSH_MASK6 );\
/* SubBytes + MixBytes */\
SUBMIX(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
\
@@ -306,16 +316,17 @@ __m512i ALL_FF;
xmm4 = _mm512_xor_si512( xmm4, xmm9 );\
xmm5 = _mm512_xor_si512( xmm5, xmm9 );\
xmm6 = _mm512_xor_si512( xmm6, xmm9 );\
xmm7 = _mm512_xor_si512( xmm7, ( ROUND_CONST_Q[ round_counter+1 ] ) );\
xmm7 = _mm512_xor_si512( xmm7, m512_const1_128( \
casti_m128i( round_const_q, round_counter+1 ) ) ); \
/* ShiftBytes Q1024 + pre-AESENCLAST */\
xmm0 = _mm512_shuffle_epi8( xmm0, ( SUBSH_MASK[1] ) );\
xmm1 = _mm512_shuffle_epi8( xmm1, ( SUBSH_MASK[3] ) );\
xmm2 = _mm512_shuffle_epi8( xmm2, ( SUBSH_MASK[5] ) );\
xmm3 = _mm512_shuffle_epi8( xmm3, ( SUBSH_MASK[7] ) );\
xmm4 = _mm512_shuffle_epi8( xmm4, ( SUBSH_MASK[0] ) );\
xmm5 = _mm512_shuffle_epi8( xmm5, ( SUBSH_MASK[2] ) );\
xmm6 = _mm512_shuffle_epi8( xmm6, ( SUBSH_MASK[4] ) );\
xmm7 = _mm512_shuffle_epi8( xmm7, ( SUBSH_MASK[6] ) );\
xmm0 = _mm512_shuffle_epi8( xmm0, SUBSH_MASK1 );\
xmm1 = _mm512_shuffle_epi8( xmm1, SUBSH_MASK3 );\
xmm2 = _mm512_shuffle_epi8( xmm2, SUBSH_MASK5 );\
xmm3 = _mm512_shuffle_epi8( xmm3, SUBSH_MASK7 );\
xmm4 = _mm512_shuffle_epi8( xmm4, SUBSH_MASK0 );\
xmm5 = _mm512_shuffle_epi8( xmm5, SUBSH_MASK2 );\
xmm6 = _mm512_shuffle_epi8( xmm6, SUBSH_MASK4 );\
xmm7 = _mm512_shuffle_epi8( xmm7, SUBSH_MASK6 );\
/* SubBytes + MixBytes */\
SUBMIX(xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
}\

4
algo/groestl/myr-groestl.c
View File

@@ -1,4 +1,7 @@
#include "myrgr-gate.h"
#if !defined(MYRGR_8WAY) && !defined(MYRGR_4WAY)
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
@@ -86,3 +89,4 @@ int scanhash_myriad( struct work *work, uint32_t max_nonce,
*hashes_done = pdata[19] - first_nonce + 1;
return 0;
}
#endif

4
algo/groestl/sph_groestl.c
View File

@@ -35,6 +35,8 @@
#include "sph_groestl.h"
#if !defined(__AES__)
#ifdef __cplusplus
extern "C"{
#endif
@@ -3116,4 +3118,6 @@ sph_groestl512_addbits_and_close(void *cc, unsigned ub, unsigned n, void *dst)
#ifdef __cplusplus
}
#endif // !AES
#endif

2
algo/groestl/sph_groestl.h
View File

@@ -42,6 +42,7 @@ extern "C"{
#include <stddef.h>
#include "algo/sha/sph_types.h"
#if !defined(__AES__)
/**
* Output size (in bits) for Groestl-224.
*/
@@ -326,4 +327,5 @@ void sph_groestl512_addbits_and_close(
}
#endif
#endif // !AES
#endif