Merge remote-tracking branch 'refs/remotes/xmrig/master'
Conflicts: CHANGELOG.md README.md src/App.cpp src/App.h src/Cpu.cpp src/Options.cpp src/Options.h src/crypto/CryptoNight_arm.h src/crypto/CryptoNight_x86.h src/version.h
This commit is contained in:
BenDroid
commit
892c7207fb
8 changed files with 76 additions and 56 deletions
|
|
@ -208,14 +208,14 @@ aes_round(__m128i key, __m128i* x0, __m128i* x1, __m128i* x2, __m128i* x3, __m12
|
|||
__m128i* x7)
|
||||
{
|
||||
if (SOFT_AES) {
|
||||
*x0 = soft_aesenc(*x0, key);
|
||||
*x1 = soft_aesenc(*x1, key);
|
||||
*x2 = soft_aesenc(*x2, key);
|
||||
*x3 = soft_aesenc(*x3, key);
|
||||
*x4 = soft_aesenc(*x4, key);
|
||||
*x5 = soft_aesenc(*x5, key);
|
||||
*x6 = soft_aesenc(*x6, key);
|
||||
*x7 = soft_aesenc(*x7, key);
|
||||
*x0 = soft_aesenc((uint32_t*)x0, key);
|
||||
*x1 = soft_aesenc((uint32_t*)x1, key);
|
||||
*x2 = soft_aesenc((uint32_t*)x2, key);
|
||||
*x3 = soft_aesenc((uint32_t*)x3, key);
|
||||
*x4 = soft_aesenc((uint32_t*)x4, key);
|
||||
*x5 = soft_aesenc((uint32_t*)x5, key);
|
||||
*x6 = soft_aesenc((uint32_t*)x6, key);
|
||||
*x7 = soft_aesenc((uint32_t*)x7, key);
|
||||
}
|
||||
# ifndef XMRIG_ARMv7
|
||||
else {
|
||||
|
|
@ -459,14 +459,16 @@ public:
|
|||
bx = _mm_set_epi64x(h[3] ^ h[7], h[2] ^ h[6]);
|
||||
idx = h[0] ^ h[4];
|
||||
|
||||
for (size_t i = 0; i < ITERATIONS; i++) {
|
||||
__m128i cx = _mm_load_si128((__m128i*) &l[idx & MASK]);
|
||||
for (size_t i = 0; i < ITERATIONS; i++) {
|
||||
__m128i cx;
|
||||
|
||||
if (SOFT_AES) {
|
||||
cx = soft_aesenc(cx, _mm_set_epi64x(ah, al));
|
||||
} else {
|
||||
# ifndef XMRIG_ARMv7
|
||||
cx = vreinterpretq_m128i_u8(vaesmcq_u8(vaeseq_u8(cx, vdupq_n_u8(0)))) ^ _mm_set_epi64x(ah, al);
|
||||
if (SOFT_AES) {
|
||||
cx = soft_aesenc((uint32_t*)&l0[idx0 & MASK], _mm_set_epi64x(ah0, al0));
|
||||
}
|
||||
else {
|
||||
cx = _mm_load_si128((__m128i *) &l0[idx0 & MASK]);
|
||||
# ifndef XMRIG_ARMv7
|
||||
cx = vreinterpretq_m128i_u8(vaesmcq_u8(vaeseq_u8(cx, vdupq_n_u8(0)))) ^ _mm_set_epi64x(ah0, al0);
|
||||
# endif
|
||||
}
|
||||
|
||||
|
|
@ -630,20 +632,19 @@ public:
|
|||
uint64_t idx1 = h1[0] ^h1[4];
|
||||
uint64_t idx2 = h2[0] ^h2[4];
|
||||
|
||||
for (size_t i = 0; i < ITERATIONS; i++) {
|
||||
__m128i cx0 = _mm_load_si128((__m128i*) &l0[idx0 & MASK]);
|
||||
__m128i cx1 = _mm_load_si128((__m128i*) &l1[idx1 & MASK]);
|
||||
__m128i cx2 = _mm_load_si128((__m128i*) &l2[idx2 & MASK]);
|
||||
for (size_t i = 0; i < ITERATIONS; i++) {
|
||||
__m128i cx0, cx1;
|
||||
|
||||
if (SOFT_AES) {
|
||||
cx0 = soft_aesenc(cx0, _mm_set_epi64x(ah0, al0));
|
||||
cx1 = soft_aesenc(cx1, _mm_set_epi64x(ah1, al1));
|
||||
cx2 = soft_aesenc(cx2, _mm_set_epi64x(ah2, al2));
|
||||
} else {
|
||||
# ifndef XMRIG_ARMv7
|
||||
cx0 = vreinterpretq_m128i_u8(vaesmcq_u8(vaeseq_u8(cx0, vdupq_n_u8(0)))) ^ _mm_set_epi64x(ah0, al0);
|
||||
cx1 = vreinterpretq_m128i_u8(vaesmcq_u8(vaeseq_u8(cx1, vdupq_n_u8(0)))) ^ _mm_set_epi64x(ah1, al1);
|
||||
cx2 = vreinterpretq_m128i_u8(vaesmcq_u8(vaeseq_u8(cx2, vdupq_n_u8(0)))) ^ _mm_set_epi64x(ah2, al2);
|
||||
if (SOFT_AES) {
|
||||
cx0 = soft_aesenc((uint32_t*)&l0[idx0 & MASK], _mm_set_epi64x(ah0, al0));
|
||||
cx1 = soft_aesenc((uint32_t*)&l1[idx1 & MASK], _mm_set_epi64x(ah1, al1));
|
||||
}
|
||||
else {
|
||||
cx0 = _mm_load_si128((__m128i *) &l0[idx0 & MASK]);
|
||||
cx1 = _mm_load_si128((__m128i *) &l1[idx1 & MASK]);
|
||||
# ifndef XMRIG_ARMv7
|
||||
cx0 = vreinterpretq_m128i_u8(vaesmcq_u8(vaeseq_u8(cx0, vdupq_n_u8(0)))) ^ _mm_set_epi64x(ah0, al0);
|
||||
cx1 = vreinterpretq_m128i_u8(vaesmcq_u8(vaeseq_u8(cx1, vdupq_n_u8(0)))) ^ _mm_set_epi64x(ah1, al1);
|
||||
# endif
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -204,15 +204,16 @@ aes_round(__m128i key, __m128i* x0, __m128i* x1, __m128i* x2, __m128i* x3, __m12
|
|||
__m128i* x7)
|
||||
{
|
||||
if (SOFT_AES) {
|
||||
*x0 = soft_aesenc(*x0, key);
|
||||
*x1 = soft_aesenc(*x1, key);
|
||||
*x2 = soft_aesenc(*x2, key);
|
||||
*x3 = soft_aesenc(*x3, key);
|
||||
*x4 = soft_aesenc(*x4, key);
|
||||
*x5 = soft_aesenc(*x5, key);
|
||||
*x6 = soft_aesenc(*x6, key);
|
||||
*x7 = soft_aesenc(*x7, key);
|
||||
} else {
|
||||
*x0 = soft_aesenc((uint32_t*)x0, key);
|
||||
*x1 = soft_aesenc((uint32_t*)x1, key);
|
||||
*x2 = soft_aesenc((uint32_t*)x2, key);
|
||||
*x3 = soft_aesenc((uint32_t*)x3, key);
|
||||
*x4 = soft_aesenc((uint32_t*)x4, key);
|
||||
*x5 = soft_aesenc((uint32_t*)x5, key);
|
||||
*x6 = soft_aesenc((uint32_t*)x6, key);
|
||||
*x7 = soft_aesenc((uint32_t*)x7, key);
|
||||
}
|
||||
else {
|
||||
*x0 = _mm_aesenc_si128(*x0, key);
|
||||
*x1 = _mm_aesenc_si128(*x1, key);
|
||||
*x2 = _mm_aesenc_si128(*x2, key);
|
||||
|
|
|
|||
|
|
@ -89,34 +89,34 @@
|
|||
alignas(16) const uint32_t saes_table[4][256] = { saes_data(saes_u0), saes_data(saes_u1), saes_data(saes_u2), saes_data(saes_u3) };
|
||||
alignas(16) const uint8_t saes_sbox[256] = saes_data(saes_h0);
|
||||
|
||||
static inline __m128i soft_aesenc(__m128i in, __m128i key)
|
||||
static inline __m128i soft_aesenc(const uint32_t* in, __m128i key)
|
||||
{
|
||||
const uint32_t x0 = _mm_cvtsi128_si32(in);
|
||||
const uint32_t x1 = _mm_cvtsi128_si32(_mm_shuffle_epi32(in, 0x55));
|
||||
const uint32_t x2 = _mm_cvtsi128_si32(_mm_shuffle_epi32(in, 0xAA));
|
||||
const uint32_t x3 = _mm_cvtsi128_si32(_mm_shuffle_epi32(in, 0xFF));
|
||||
const uint32_t x0 = in[0];
|
||||
const uint32_t x1 = in[1];
|
||||
const uint32_t x2 = in[2];
|
||||
const uint32_t x3 = in[3];
|
||||
|
||||
__m128i out = _mm_set_epi32(
|
||||
(saes_table[0][x3 & 0xff] ^ saes_table[1][(x0 >> 8) & 0xff] ^ saes_table[2][(x1 >> 16) & 0xff] ^ saes_table[3][x2 >> 24]),
|
||||
(saes_table[0][x2 & 0xff] ^ saes_table[1][(x3 >> 8) & 0xff] ^ saes_table[2][(x0 >> 16) & 0xff] ^ saes_table[3][x1 >> 24]),
|
||||
(saes_table[0][x1 & 0xff] ^ saes_table[1][(x2 >> 8) & 0xff] ^ saes_table[2][(x3 >> 16) & 0xff] ^ saes_table[3][x0 >> 24]),
|
||||
(saes_table[0][x0 & 0xff] ^ saes_table[1][(x1 >> 8) & 0xff] ^ saes_table[2][(x2 >> 16) & 0xff] ^ saes_table[3][x3 >> 24]));
|
||||
__m128i out = _mm_set_epi32(
|
||||
(saes_table[0][x3 & 0xff] ^ saes_table[1][(x0 >> 8) & 0xff] ^ saes_table[2][(x1 >> 16) & 0xff] ^ saes_table[3][x2 >> 24]),
|
||||
(saes_table[0][x2 & 0xff] ^ saes_table[1][(x3 >> 8) & 0xff] ^ saes_table[2][(x0 >> 16) & 0xff] ^ saes_table[3][x1 >> 24]),
|
||||
(saes_table[0][x1 & 0xff] ^ saes_table[1][(x2 >> 8) & 0xff] ^ saes_table[2][(x3 >> 16) & 0xff] ^ saes_table[3][x0 >> 24]),
|
||||
(saes_table[0][x0 & 0xff] ^ saes_table[1][(x1 >> 8) & 0xff] ^ saes_table[2][(x2 >> 16) & 0xff] ^ saes_table[3][x3 >> 24]));
|
||||
|
||||
return _mm_xor_si128(out, key);
|
||||
return _mm_xor_si128(out, key);
|
||||
}
|
||||
|
||||
static inline uint32_t sub_word(uint32_t key)
|
||||
{
|
||||
return (saes_sbox[key >> 24 ] << 24) |
|
||||
(saes_sbox[(key >> 16) & 0xff] << 16 ) |
|
||||
(saes_sbox[(key >> 8) & 0xff] << 8 ) |
|
||||
saes_sbox[key & 0xff];
|
||||
return (saes_sbox[key >> 24 ] << 24) |
|
||||
(saes_sbox[(key >> 16) & 0xff] << 16 ) |
|
||||
(saes_sbox[(key >> 8) & 0xff] << 8 ) |
|
||||
saes_sbox[key & 0xff];
|
||||
}
|
||||
|
||||
#if defined(__clang__) || defined(XMRIG_ARM)
|
||||
static inline uint32_t _rotr(uint32_t value, uint32_t amount)
|
||||
{
|
||||
return (value >> amount) | (value << ((32 - amount) & 31));
|
||||
return (value >> amount) | (value << ((32 - amount) & 31));
|
||||
}
|
||||
#endif
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue