mirror of
https://github.com/xmrig/xmrig.git
synced 2025-12-25 05:40:39 -05:00
Merge xmrig v6.16.0 into master
This commit is contained in:
MoneroOcean
@@ -43,6 +43,7 @@ public:
|
||||
constexpr inline size_t memory() const { static_assert(Algorithm::isCN(ALGO), "invalid CRYPTONIGHT algorithm"); return Algorithm::l3(ALGO); }
|
||||
constexpr inline uint32_t iterations() const { static_assert(Algorithm::isCN(ALGO), "invalid CRYPTONIGHT algorithm"); return CN_ITER; }
|
||||
constexpr inline uint32_t mask() const { return static_cast<uint32_t>(((memory() - 1) / 16) * 16); }
|
||||
constexpr inline uint32_t half_mem() const { return mask() < memory() / 2; }
|
||||
|
||||
inline static uint32_t iterations(Algorithm::Id algo)
|
||||
{
|
||||
@@ -119,6 +120,16 @@ public:
|
||||
}
|
||||
# endif
|
||||
|
||||
# ifdef XMRIG_ALGO_GHOSTRIDER
|
||||
if (algo == Algorithm::CN_GR_1) {
|
||||
return 0x3FFF0;
|
||||
}
|
||||
|
||||
if (algo == Algorithm::CN_GR_5) {
|
||||
return 0x1FFF0;
|
||||
}
|
||||
# endif
|
||||
|
||||
return ((Algorithm::l3(algo) - 1) / 16) * 16;
|
||||
}
|
||||
|
||||
@@ -149,6 +160,18 @@ template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_PICO_0>::mask() const
|
||||
template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_GPU>::mask() const { return 0x1FFFC0; }
|
||||
template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_UPX2>::mask() const { return 0x1FFF0; }
|
||||
|
||||
#ifdef XMRIG_ALGO_GHOSTRIDER
|
||||
template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_GR_0>::iterations() const { return CN_ITER / 4; }
|
||||
template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_GR_1>::iterations() const { return CN_ITER / 4; }
|
||||
template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_GR_2>::iterations() const { return CN_ITER / 2; }
|
||||
template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_GR_3>::iterations() const { return CN_ITER / 2; }
|
||||
template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_GR_4>::iterations() const { return CN_ITER / 8; }
|
||||
template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_GR_5>::iterations() const { return CN_ITER / 8; }
|
||||
|
||||
template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_GR_1>::mask() const { return 0x3FFF0; }
|
||||
template<> constexpr inline uint32_t CnAlgo<Algorithm::CN_GR_5>::mask() const { return 0x1FFF0; }
|
||||
#endif
|
||||
|
||||
|
||||
} /* namespace xmrig */
|
||||
|
||||
|
||||
@@ -316,6 +316,15 @@ xmrig::CnHash::CnHash()
|
||||
m_map[Algorithm::CN_GPU]->data[AV_SINGLE_SOFT][Assembly::NONE] = cryptonight_single_hash_gpu<Algorithm::CN_GPU, true>;
|
||||
# endif
|
||||
|
||||
# ifdef XMRIG_ALGO_GHOSTRIDER
|
||||
ADD_FN(Algorithm::CN_GR_0);
|
||||
ADD_FN(Algorithm::CN_GR_1);
|
||||
ADD_FN(Algorithm::CN_GR_2);
|
||||
ADD_FN(Algorithm::CN_GR_3);
|
||||
ADD_FN(Algorithm::CN_GR_4);
|
||||
ADD_FN(Algorithm::CN_GR_5);
|
||||
# endif
|
||||
|
||||
# ifdef XMRIG_FEATURE_ASM
|
||||
patchAsmVariants();
|
||||
# endif
|
||||
|
||||
@@ -58,6 +58,9 @@ struct cryptonight_ctx {
|
||||
|
||||
cn_mainloop_fun_ms_abi generated_code;
|
||||
cryptonight_r_data generated_code_data;
|
||||
|
||||
alignas(16) uint8_t save_state[128];
|
||||
bool first_half;
|
||||
};
|
||||
|
||||
|
||||
|
||||
@@ -353,6 +353,9 @@ static inline __m128i aes_round_tweak_div(const __m128i &in, const __m128i &key)
|
||||
}
|
||||
|
||||
|
||||
alignas(64) static const uint32_t tweak1_table[256] = { 268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456 };
|
||||
|
||||
|
||||
namespace xmrig {
|
||||
|
||||
|
||||
@@ -372,12 +375,7 @@ static inline void cryptonight_monero_tweak(const uint8_t* l, uint64_t idx, __m1
|
||||
|
||||
uint64_t vh = vgetq_lane_u64(tmp, 1);
|
||||
|
||||
uint8_t x = vh >> 24;
|
||||
static const uint16_t table = 0x7531;
|
||||
const uint8_t index = (((x >> (3)) & 6) | (x & 1)) << 1;
|
||||
vh ^= ((table >> index) & 0x3) << 28;
|
||||
|
||||
mem_out[1] = vh;
|
||||
mem_out[1] = vh ^ tweak1_table[static_cast<uint8_t>(vh >> 24)];
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@@ -100,7 +100,7 @@ const static uint8_t test_output_r[] = {
|
||||
|
||||
|
||||
// "cn/0"
|
||||
const static uint8_t test_output_v0[160] = {
|
||||
const static uint8_t test_output_v0[256] = {
|
||||
0x1A, 0x3F, 0xFB, 0xEE, 0x90, 0x9B, 0x42, 0x0D, 0x91, 0xF7, 0xBE, 0x6E, 0x5F, 0xB5, 0x6D, 0xB7,
|
||||
0x1B, 0x31, 0x10, 0xD8, 0x86, 0x01, 0x1E, 0x87, 0x7E, 0xE5, 0x78, 0x6A, 0xFD, 0x08, 0x01, 0x00,
|
||||
0x1B, 0x60, 0x6A, 0x3F, 0x4A, 0x07, 0xD6, 0x48, 0x9A, 0x1B, 0xCD, 0x07, 0x69, 0x7B, 0xD1, 0x66,
|
||||
@@ -115,7 +115,7 @@ const static uint8_t test_output_v0[160] = {
|
||||
|
||||
|
||||
// "cn/1" Cryptonight variant 1 (Monero v7)
|
||||
const static uint8_t test_output_v1[160] = {
|
||||
const static uint8_t test_output_v1[256] = {
|
||||
0xF2, 0x2D, 0x3D, 0x62, 0x03, 0xD2, 0xA0, 0x8B, 0x41, 0xD9, 0x02, 0x72, 0x78, 0xD8, 0xBC, 0xC9,
|
||||
0x83, 0xAC, 0xAD, 0xA9, 0xB6, 0x8E, 0x52, 0xE3, 0xC6, 0x89, 0x69, 0x2A, 0x50, 0xE9, 0x21, 0xD9,
|
||||
0xC9, 0xFA, 0xE8, 0x42, 0x5D, 0x86, 0x88, 0xDC, 0x23, 0x6B, 0xCD, 0xBC, 0x42, 0xFD, 0xB4, 0x2D,
|
||||
@@ -130,7 +130,7 @@ const static uint8_t test_output_v1[160] = {
|
||||
|
||||
|
||||
// "cn/2" Cryptonight variant 2 (Monero v8)
|
||||
const static uint8_t test_output_v2[160] = {
|
||||
const static uint8_t test_output_v2[256] = {
|
||||
0x97, 0x37, 0x82, 0x82, 0xCF, 0x10, 0xE7, 0xAD, 0x03, 0x3F, 0x7B, 0x80, 0x74, 0xC4, 0x0E, 0x14,
|
||||
0xD0, 0x6E, 0x7F, 0x60, 0x9D, 0xDD, 0xDA, 0x78, 0x76, 0x80, 0xB5, 0x8C, 0x05, 0xF4, 0x3D, 0x21,
|
||||
0x87, 0x1F, 0xCD, 0x68, 0x23, 0xF6, 0xA8, 0x79, 0xBB, 0x3F, 0x33, 0x95, 0x1C, 0x8E, 0x8E, 0x89,
|
||||
@@ -145,7 +145,7 @@ const static uint8_t test_output_v2[160] = {
|
||||
|
||||
|
||||
// "cn/half"
|
||||
const static uint8_t test_output_half[160] = {
|
||||
const static uint8_t test_output_half[256] = {
|
||||
0x5D, 0x4F, 0xBC, 0x35, 0x60, 0x97, 0xEA, 0x64, 0x40, 0xB0, 0x88, 0x8E, 0xDE, 0xB6, 0x35, 0xDD,
|
||||
0xC8, 0x4A, 0x0E, 0x39, 0x7C, 0x86, 0x84, 0x56, 0x89, 0x5C, 0x3F, 0x29, 0xBE, 0x73, 0x12, 0xA7,
|
||||
0x02, 0xE6, 0x1D, 0x2B, 0xBC, 0x84, 0xB6, 0x71, 0x96, 0x71, 0xD5, 0x0C, 0xAC, 0x76, 0x0E, 0x6B,
|
||||
@@ -160,7 +160,7 @@ const static uint8_t test_output_half[160] = {
|
||||
|
||||
|
||||
// "cn/msr" Masari (MSR)
|
||||
const static uint8_t test_output_msr[160] = {
|
||||
const static uint8_t test_output_msr[256] = {
|
||||
0x3C, 0x7A, 0x61, 0x08, 0x4C, 0x5E, 0xB8, 0x65, 0xB4, 0x98, 0xAB, 0x2F, 0x5A, 0x1A, 0xC5, 0x2C,
|
||||
0x49, 0xC1, 0x77, 0xC2, 0xD0, 0x13, 0x34, 0x42, 0xD6, 0x5E, 0xD5, 0x14, 0x33, 0x5C, 0x82, 0xC5,
|
||||
0x69, 0xDF, 0x38, 0x51, 0x1B, 0xB3, 0xEB, 0x7D, 0xE7, 0x6B, 0x08, 0x8E, 0xB6, 0x7E, 0xB7, 0x1C,
|
||||
@@ -175,7 +175,7 @@ const static uint8_t test_output_msr[160] = {
|
||||
|
||||
|
||||
// "cn/xao" Alloy (XAO)
|
||||
const static uint8_t test_output_xao[160] = {
|
||||
const static uint8_t test_output_xao[256] = {
|
||||
0x9A, 0x29, 0xD0, 0xC4, 0xAF, 0xDC, 0x63, 0x9B, 0x65, 0x53, 0xB1, 0xC8, 0x37, 0x35, 0x11, 0x4C,
|
||||
0x5D, 0x77, 0x16, 0x21, 0x42, 0x97, 0x5C, 0xB8, 0x50, 0xC0, 0xA5, 0x1F, 0x64, 0x07, 0xBD, 0x33,
|
||||
0xF1, 0xC9, 0x98, 0x40, 0x42, 0xDE, 0x39, 0xD1, 0xBA, 0x2D, 0xAD, 0xEC, 0xFE, 0xEA, 0xD8, 0x46,
|
||||
@@ -190,7 +190,7 @@ const static uint8_t test_output_xao[160] = {
|
||||
|
||||
|
||||
// "cn/rto" Arto (RTO)
|
||||
const static uint8_t test_output_rto[160] = {
|
||||
const static uint8_t test_output_rto[256] = {
|
||||
0x82, 0x66, 0x1E, 0x1C, 0x6E, 0x64, 0x36, 0x66, 0x84, 0x06, 0x32, 0x7A, 0x9B, 0xB1, 0x13, 0x19,
|
||||
0xA5, 0x56, 0x16, 0x15, 0xDF, 0xEC, 0x1C, 0x9E, 0xE3, 0x88, 0x4A, 0x6C, 0x1C, 0xEB, 0x76, 0xA5,
|
||||
0xB3, 0xFB, 0xF4, 0x3F, 0x2B, 0x6A, 0x3A, 0x39, 0xA3, 0x6E, 0x08, 0x33, 0x67, 0x90, 0x31, 0xB9,
|
||||
@@ -204,7 +204,7 @@ const static uint8_t test_output_rto[160] = {
|
||||
};
|
||||
|
||||
// "cn/rwz"
|
||||
const static uint8_t test_output_rwz[160] = {
|
||||
const static uint8_t test_output_rwz[256] = {
|
||||
0x5f, 0x56, 0xc6, 0xb0, 0x99, 0x6b, 0xa2, 0x3e, 0x0b, 0xba, 0x07, 0x29, 0xc9, 0x90, 0x74, 0x85,
|
||||
0x5a, 0x10, 0xe3, 0x08, 0x7f, 0xdb, 0xfe, 0x94, 0x75, 0x33, 0x54, 0x73, 0x76, 0xf0, 0x75, 0xb8,
|
||||
0x8b, 0x70, 0x43, 0x9a, 0xfc, 0xf5, 0xeb, 0x15, 0xbb, 0xf9, 0xad, 0x9d, 0x2a, 0xbd, 0x72, 0x52,
|
||||
@@ -218,7 +218,7 @@ const static uint8_t test_output_rwz[160] = {
|
||||
};
|
||||
|
||||
// "cn/zls"
|
||||
const static uint8_t test_output_zls[160] = {
|
||||
const static uint8_t test_output_zls[256] = {
|
||||
0x51, 0x6E, 0x33, 0xC6, 0xE4, 0x46, 0xAB, 0xBC, 0xCD, 0xAD, 0x18, 0xC0, 0x4C, 0xD9, 0xA2, 0x5E,
|
||||
0x64, 0x10, 0x28, 0x53, 0xB2, 0x0A, 0x42, 0xDF, 0xDE, 0xAA, 0x8B, 0x59, 0x9E, 0xCF, 0x40, 0xE2,
|
||||
0x0D, 0x62, 0x5B, 0x42, 0x18, 0xE2, 0x76, 0xAD, 0xD0, 0x74, 0x90, 0x60, 0x8D, 0xC4, 0xC7, 0x80,
|
||||
@@ -232,7 +232,7 @@ const static uint8_t test_output_zls[160] = {
|
||||
};
|
||||
|
||||
// "cn/ccx"
|
||||
const static uint8_t test_output_ccx[160] = {
|
||||
const static uint8_t test_output_ccx[256] = {
|
||||
0xB3, 0xA1, 0x67, 0x86, 0xD2, 0xC9, 0x85, 0xEC, 0xAD, 0xC4, 0x5F, 0x91, 0x05, 0x27, 0xC7, 0xA1,
|
||||
0x96, 0xF0, 0xE1, 0xE9, 0x7C, 0x87, 0x09, 0x38, 0x1D, 0x7D, 0x41, 0x93, 0x35, 0xF8, 0x16, 0x72,
|
||||
0xC3, 0xBD, 0x8D, 0xE8, 0xD5, 0xAE, 0xB8, 0x59, 0x0A, 0x6C, 0xCB, 0x7B, 0x41, 0x30, 0xF7, 0x04,
|
||||
@@ -246,7 +246,7 @@ const static uint8_t test_output_ccx[160] = {
|
||||
};
|
||||
|
||||
// "cn/double"
|
||||
const static uint8_t test_output_double[160] = {
|
||||
const static uint8_t test_output_double[256] = {
|
||||
0xAE, 0xFB, 0xB3, 0xF0, 0xCC, 0x88, 0x04, 0x6D, 0x11, 0x9F, 0x6C, 0x54, 0xB9, 0x6D, 0x90, 0xC9,
|
||||
0xE8, 0x84, 0xEA, 0x3B, 0x59, 0x83, 0xA6, 0x0D, 0x50, 0xA4, 0x2D, 0x7D, 0x3E, 0xBE, 0x48, 0x21,
|
||||
0x49, 0xCE, 0x8E, 0xF3, 0xBC, 0x8A, 0x36, 0xBF, 0x86, 0x37, 0x89, 0x55, 0x09, 0xBA, 0x22, 0xF8,
|
||||
@@ -261,7 +261,7 @@ const static uint8_t test_output_double[160] = {
|
||||
|
||||
#ifdef XMRIG_ALGO_CN_LITE
|
||||
// "cn-lite/0"
|
||||
const static uint8_t test_output_v0_lite[160] = {
|
||||
const static uint8_t test_output_v0_lite[256] = {
|
||||
0x36, 0x95, 0xB4, 0xB5, 0x3B, 0xB0, 0x03, 0x58, 0xB0, 0xAD, 0x38, 0xDC, 0x16, 0x0F, 0xEB, 0x9E,
|
||||
0x00, 0x4E, 0xEC, 0xE0, 0x9B, 0x83, 0xA7, 0x2E, 0xF6, 0xBA, 0x98, 0x64, 0xD3, 0x51, 0x0C, 0x88,
|
||||
0x28, 0xA2, 0x2B, 0xAD, 0x3F, 0x93, 0xD1, 0x40, 0x8F, 0xCA, 0x47, 0x2E, 0xB5, 0xAD, 0x1C, 0xBE,
|
||||
@@ -276,7 +276,7 @@ const static uint8_t test_output_v0_lite[160] = {
|
||||
|
||||
|
||||
// "cn-lite/1" AEON v7
|
||||
const static uint8_t test_output_v1_lite[160] = {
|
||||
const static uint8_t test_output_v1_lite[256] = {
|
||||
0x6D, 0x8C, 0xDC, 0x44, 0x4E, 0x9B, 0xBB, 0xFD, 0x68, 0xFC, 0x43, 0xFC, 0xD4, 0x85, 0x5B, 0x22,
|
||||
0x8C, 0x8A, 0x1B, 0xD9, 0x1D, 0x9D, 0x00, 0x28, 0x5B, 0xEC, 0x02, 0xB7, 0xCA, 0x2D, 0x67, 0x41,
|
||||
0x87, 0xC4, 0xE5, 0x70, 0x65, 0x3E, 0xB4, 0xC2, 0xB4, 0x2B, 0x7A, 0x0D, 0x54, 0x65, 0x59, 0x45,
|
||||
@@ -293,7 +293,7 @@ const static uint8_t test_output_v1_lite[160] = {
|
||||
|
||||
#ifdef XMRIG_ALGO_CN_HEAVY
|
||||
// "cn-heavy/0"
|
||||
const static uint8_t test_output_v0_heavy[160] = {
|
||||
const static uint8_t test_output_v0_heavy[256] = {
|
||||
0x99, 0x83, 0xF2, 0x1B, 0xDF, 0x20, 0x10, 0xA8, 0xD7, 0x07, 0xBB, 0x2F, 0x14, 0xD7, 0x86, 0x64,
|
||||
0xBB, 0xE1, 0x18, 0x7F, 0x55, 0x01, 0x4B, 0x39, 0xE5, 0xF3, 0xD6, 0x93, 0x28, 0xE4, 0x8F, 0xC2,
|
||||
0x4D, 0x94, 0x7D, 0xD6, 0xDB, 0x6E, 0x07, 0x48, 0x26, 0x4A, 0x51, 0x2E, 0xAC, 0xF3, 0x25, 0x4A,
|
||||
@@ -308,7 +308,7 @@ const static uint8_t test_output_v0_heavy[160] = {
|
||||
|
||||
|
||||
// "cn-heavy/xhv"
|
||||
const static uint8_t test_output_xhv_heavy[160] = {
|
||||
const static uint8_t test_output_xhv_heavy[256] = {
|
||||
0x5A, 0xC3, 0xF7, 0x85, 0xC4, 0x90, 0xC5, 0x85, 0x50, 0xEC, 0x95, 0xD2, 0x72, 0x65, 0x63, 0x57,
|
||||
0x7E, 0x7C, 0x1C, 0x21, 0x2D, 0x0C, 0xDE, 0x59, 0x12, 0x73, 0x20, 0x1E, 0x44, 0xFD, 0xD5, 0xB6,
|
||||
0x1F, 0x4E, 0xB2, 0x0A, 0x36, 0x51, 0x4B, 0xF5, 0x4D, 0xC9, 0xE0, 0x90, 0x2C, 0x16, 0x47, 0x3F,
|
||||
@@ -323,7 +323,7 @@ const static uint8_t test_output_xhv_heavy[160] = {
|
||||
|
||||
|
||||
// "cn-heavy/tube"
|
||||
const static uint8_t test_output_tube_heavy[160] = {
|
||||
const static uint8_t test_output_tube_heavy[256] = {
|
||||
0xFE, 0x53, 0x35, 0x20, 0x76, 0xEA, 0xE6, 0x89, 0xFA, 0x3B, 0x4F, 0xDA, 0x61, 0x46, 0x34, 0xCF,
|
||||
0xC3, 0x12, 0xEE, 0x0C, 0x38, 0x7D, 0xF2, 0xB8, 0xB7, 0x4D, 0xA2, 0xA1, 0x59, 0x74, 0x12, 0x35,
|
||||
0xCD, 0x3F, 0x29, 0xDF, 0x07, 0x4A, 0x14, 0xAD, 0x0B, 0x98, 0x99, 0x37, 0xCA, 0x14, 0x68, 0xA3,
|
||||
@@ -340,7 +340,7 @@ const static uint8_t test_output_tube_heavy[160] = {
|
||||
|
||||
#ifdef XMRIG_ALGO_CN_PICO
|
||||
// "cn-pico/trtl"
|
||||
const static uint8_t test_output_pico_trtl[160] = {
|
||||
const static uint8_t test_output_pico_trtl[256] = {
|
||||
0x08, 0xF4, 0x21, 0xD7, 0x83, 0x31, 0x17, 0x30, 0x0E, 0xDA, 0x66, 0xE9, 0x8F, 0x4A, 0x25, 0x69,
|
||||
0x09, 0x3D, 0xF3, 0x00, 0x50, 0x01, 0x73, 0x94, 0x4E, 0xFC, 0x40, 0x1E, 0x9A, 0x4A, 0x17, 0xAF,
|
||||
0xB2, 0x17, 0x2E, 0xC9, 0x46, 0x6E, 0x1A, 0xEE, 0x70, 0xEC, 0x85, 0x72, 0xA1, 0x4C, 0x23, 0x3E,
|
||||
@@ -355,7 +355,7 @@ const static uint8_t test_output_pico_trtl[160] = {
|
||||
|
||||
|
||||
// "cn-pico/tlo"
|
||||
const static uint8_t test_output_pico_tlo[160] = {
|
||||
const static uint8_t test_output_pico_tlo[256] = {
|
||||
0x99, 0x75, 0xF2, 0xC1, 0xB3, 0xB4, 0x54, 0x34, 0xA4, 0x93, 0x86, 0x21, 0x30, 0x97, 0xF3, 0x1B,
|
||||
0xB4, 0xB9, 0xA6, 0x58, 0x6A, 0x7E, 0x81, 0xF4, 0x42, 0x9F, 0x6D, 0x5F, 0x65, 0xC3, 0x8D, 0x1A,
|
||||
0xFC, 0x67, 0xDF, 0xCC, 0xB5, 0xFC, 0x90, 0xD7, 0x85, 0x5A, 0xE9, 0x03, 0x36, 0x1E, 0xAB, 0xD7,
|
||||
@@ -388,7 +388,7 @@ const static uint8_t test_output_gpu[160] = {
|
||||
|
||||
#ifdef XMRIG_ALGO_CN_FEMTO
|
||||
// "cn/upx2"
|
||||
const static uint8_t test_output_femto_upx2[160] = {
|
||||
const static uint8_t test_output_femto_upx2[256] = {
|
||||
0xAA, 0xBB, 0xB8, 0xED, 0x14, 0xA8, 0x35, 0xFA, 0x22, 0xCF, 0xB1, 0xB5, 0xDE, 0xA8, 0x72, 0xB0,
|
||||
0xA1, 0xD6, 0xCB, 0xD8, 0x46, 0xF4, 0x39, 0x1C, 0x0F, 0x01, 0xF3, 0x87, 0x5E, 0x3A, 0x37, 0x61,
|
||||
0x38, 0x59, 0x15, 0x72, 0xF8, 0x20, 0xD4, 0xDE, 0x25, 0x3C, 0xF5, 0x5A, 0x21, 0x92, 0xB6, 0x22,
|
||||
@@ -405,7 +405,7 @@ const static uint8_t test_output_femto_upx2[160] = {
|
||||
|
||||
#ifdef XMRIG_ALGO_ARGON2
|
||||
// "argon2/chukwa"
|
||||
const static uint8_t argon2_chukwa_test_out[160] = {
|
||||
const static uint8_t argon2_chukwa_test_out[256] = {
|
||||
0xC1, 0x58, 0xA1, 0x05, 0xAE, 0x75, 0xC7, 0x56, 0x1C, 0xFD, 0x02, 0x90, 0x83, 0xA4, 0x7A, 0x87,
|
||||
0x65, 0x3D, 0x51, 0xF9, 0x14, 0x12, 0x8E, 0x21, 0xC1, 0x97, 0x1D, 0x8B, 0x10, 0xC4, 0x90, 0x34,
|
||||
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
|
||||
@@ -419,7 +419,7 @@ const static uint8_t argon2_chukwa_test_out[160] = {
|
||||
};
|
||||
|
||||
// "argon2/chukwav2"
|
||||
const static uint8_t argon2_chukwa_v2_test_out[160] = {
|
||||
const static uint8_t argon2_chukwa_v2_test_out[256] = {
|
||||
0x77, 0xCF, 0x69, 0x58, 0xB3, 0x53, 0x6E, 0x1F, 0x9F, 0x0D, 0x1E, 0xA1, 0x65, 0xF2, 0x28, 0x11,
|
||||
0xCA, 0x7B, 0xC4, 0x87, 0xEA, 0x9F, 0x52, 0x03, 0x0B, 0x50, 0x50, 0xC1, 0x7F, 0xCD, 0xD8, 0xF5,
|
||||
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
|
||||
@@ -433,7 +433,7 @@ const static uint8_t argon2_chukwa_v2_test_out[160] = {
|
||||
};
|
||||
|
||||
// "argon2/wrkz"
|
||||
const static uint8_t argon2_wrkz_test_out[160] = {
|
||||
const static uint8_t argon2_wrkz_test_out[256] = {
|
||||
0x35, 0xE0, 0x83, 0xD4, 0xB9, 0xC6, 0x4C, 0x2A, 0x68, 0x82, 0x0A, 0x43, 0x1F, 0x61, 0x31, 0x19,
|
||||
0x98, 0xA8, 0xCD, 0x18, 0x64, 0xDB, 0xA4, 0x07, 0x7E, 0x25, 0xB7, 0xF1, 0x21, 0xD5, 0x4B, 0xD1,
|
||||
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
|
||||
@@ -450,7 +450,7 @@ const static uint8_t argon2_wrkz_test_out[160] = {
|
||||
|
||||
#ifdef XMRIG_ALGO_ASTROBWT
|
||||
// "astrobwt"
|
||||
const static uint8_t astrobwt_dero_test_out[160] = {
|
||||
const static uint8_t astrobwt_dero_test_out[256] = {
|
||||
0x7E, 0x88, 0x44, 0xF2, 0xD6, 0xB7, 0xA4, 0x34, 0x98, 0xFE, 0x6D, 0x22, 0x65, 0x27, 0x68, 0x90,
|
||||
0x23, 0xDA, 0x8A, 0x52, 0xF9, 0xFC, 0x4E, 0xC6, 0x9E, 0x5A, 0xAA, 0xA6, 0x3E, 0xDC, 0xE1, 0xC1,
|
||||
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
|
||||
|
||||
@@ -285,23 +285,41 @@ inline constexpr uint64_t interleaved_index<0>(uint64_t k)
|
||||
|
||||
|
||||
template<Algorithm::Id ALGO, bool SOFT_AES, int interleave>
|
||||
static inline void cn_explode_scratchpad(const __m128i *input, __m128i *output)
|
||||
static NOINLINE void cn_explode_scratchpad(cryptonight_ctx *ctx)
|
||||
{
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
|
||||
constexpr size_t N = (props.memory() / sizeof(__m128i)) / (props.half_mem() ? 2 : 1);
|
||||
|
||||
__m128i xin0, xin1, xin2, xin3, xin4, xin5, xin6, xin7;
|
||||
__m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9;
|
||||
|
||||
const __m128i* input = reinterpret_cast<const __m128i*>(ctx->state);
|
||||
__m128i* output = reinterpret_cast<__m128i*>(ctx->memory);
|
||||
|
||||
aes_genkey<SOFT_AES>(input, &k0, &k1, &k2, &k3, &k4, &k5, &k6, &k7, &k8, &k9);
|
||||
|
||||
xin0 = _mm_load_si128(input + 4);
|
||||
xin1 = _mm_load_si128(input + 5);
|
||||
xin2 = _mm_load_si128(input + 6);
|
||||
xin3 = _mm_load_si128(input + 7);
|
||||
xin4 = _mm_load_si128(input + 8);
|
||||
xin5 = _mm_load_si128(input + 9);
|
||||
xin6 = _mm_load_si128(input + 10);
|
||||
xin7 = _mm_load_si128(input + 11);
|
||||
if (props.half_mem() && !ctx->first_half) {
|
||||
const __m128i* p = reinterpret_cast<const __m128i*>(ctx->save_state);
|
||||
xin0 = _mm_load_si128(p + 0);
|
||||
xin1 = _mm_load_si128(p + 1);
|
||||
xin2 = _mm_load_si128(p + 2);
|
||||
xin3 = _mm_load_si128(p + 3);
|
||||
xin4 = _mm_load_si128(p + 4);
|
||||
xin5 = _mm_load_si128(p + 5);
|
||||
xin6 = _mm_load_si128(p + 6);
|
||||
xin7 = _mm_load_si128(p + 7);
|
||||
}
|
||||
else {
|
||||
xin0 = _mm_load_si128(input + 4);
|
||||
xin1 = _mm_load_si128(input + 5);
|
||||
xin2 = _mm_load_si128(input + 6);
|
||||
xin3 = _mm_load_si128(input + 7);
|
||||
xin4 = _mm_load_si128(input + 8);
|
||||
xin5 = _mm_load_si128(input + 9);
|
||||
xin6 = _mm_load_si128(input + 10);
|
||||
xin7 = _mm_load_si128(input + 11);
|
||||
}
|
||||
|
||||
if (props.isHeavy()) {
|
||||
for (size_t i = 0; i < 16; i++) {
|
||||
@@ -320,42 +338,61 @@ static inline void cn_explode_scratchpad(const __m128i *input, __m128i *output)
|
||||
}
|
||||
}
|
||||
|
||||
for (size_t i = 0; i < props.memory() / sizeof(__m128i); i += 8) {
|
||||
if (interleave > 0) {
|
||||
_mm_prefetch((const char*)(output), _MM_HINT_T0);
|
||||
_mm_prefetch((const char*)(output + (64 << interleave) / sizeof(__m128i)), _MM_HINT_T0);
|
||||
}
|
||||
constexpr int output_increment = (64 << interleave) / sizeof(__m128i);
|
||||
constexpr int prefetch_dist = 2048 / sizeof(__m128i);
|
||||
|
||||
aes_round<SOFT_AES>(k0, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k1, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k2, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k3, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k4, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k5, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k6, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k7, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k8, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k9, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
__m128i* e = output + N - prefetch_dist;
|
||||
__m128i* prefetch_ptr = output + prefetch_dist;
|
||||
|
||||
_mm_store_si128(output + 0, xin0);
|
||||
_mm_store_si128(output + 1, xin1);
|
||||
_mm_store_si128(output + 2, xin2);
|
||||
_mm_store_si128(output + 3, xin3);
|
||||
for (int i = 0; i < 2; ++i) {
|
||||
do {
|
||||
_mm_prefetch((const char*)(prefetch_ptr), _MM_HINT_T0);
|
||||
_mm_prefetch((const char*)(prefetch_ptr + output_increment), _MM_HINT_T0);
|
||||
|
||||
constexpr int output_increment = (64 << interleave) / sizeof(__m128i);
|
||||
aes_round<SOFT_AES>(k0, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k1, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k2, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k3, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k4, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k5, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k6, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k7, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k8, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
aes_round<SOFT_AES>(k9, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
|
||||
|
||||
_mm_store_si128(output + output_increment + 0, xin4);
|
||||
_mm_store_si128(output + output_increment + 1, xin5);
|
||||
_mm_store_si128(output + output_increment + 2, xin6);
|
||||
_mm_store_si128(output + output_increment + 3, xin7);
|
||||
_mm_store_si128(output + 0, xin0);
|
||||
_mm_store_si128(output + 1, xin1);
|
||||
_mm_store_si128(output + 2, xin2);
|
||||
_mm_store_si128(output + 3, xin3);
|
||||
|
||||
output += output_increment * 2;
|
||||
_mm_store_si128(output + output_increment + 0, xin4);
|
||||
_mm_store_si128(output + output_increment + 1, xin5);
|
||||
_mm_store_si128(output + output_increment + 2, xin6);
|
||||
_mm_store_si128(output + output_increment + 3, xin7);
|
||||
|
||||
output += output_increment * 2;
|
||||
prefetch_ptr += output_increment * 2;
|
||||
} while (output < e);
|
||||
e += prefetch_dist;
|
||||
prefetch_ptr = output;
|
||||
}
|
||||
|
||||
if (props.half_mem() && ctx->first_half) {
|
||||
__m128i* p = reinterpret_cast<__m128i*>(ctx->save_state);
|
||||
_mm_store_si128(p + 0, xin0);
|
||||
_mm_store_si128(p + 1, xin1);
|
||||
_mm_store_si128(p + 2, xin2);
|
||||
_mm_store_si128(p + 3, xin3);
|
||||
_mm_store_si128(p + 4, xin4);
|
||||
_mm_store_si128(p + 5, xin5);
|
||||
_mm_store_si128(p + 6, xin6);
|
||||
_mm_store_si128(p + 7, xin7);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
template<Algorithm::Id ALGO, bool SOFT_AES, int interleave>
|
||||
static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output)
|
||||
static NOINLINE void cn_implode_scratchpad(cryptonight_ctx *ctx)
|
||||
{
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
|
||||
@@ -365,9 +402,14 @@ static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output)
|
||||
constexpr bool IS_HEAVY = props.isHeavy();
|
||||
# endif
|
||||
|
||||
constexpr size_t N = (props.memory() / sizeof(__m128i)) / (props.half_mem() ? 2 : 1);
|
||||
|
||||
__m128i xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7;
|
||||
__m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9;
|
||||
|
||||
const __m128i *input = reinterpret_cast<const __m128i*>(ctx->memory);
|
||||
__m128i *output = reinterpret_cast<__m128i*>(ctx->state);
|
||||
|
||||
aes_genkey<SOFT_AES>(output + 2, &k0, &k1, &k2, &k3, &k4, &k5, &k6, &k7, &k8, &k9);
|
||||
|
||||
xout0 = _mm_load_si128(output + 4);
|
||||
@@ -380,46 +422,54 @@ static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output)
|
||||
xout7 = _mm_load_si128(output + 11);
|
||||
|
||||
const __m128i* input_begin = input;
|
||||
for (size_t i = 0; i < props.memory() / sizeof(__m128i);) {
|
||||
xout0 = _mm_xor_si128(_mm_load_si128(input + 0), xout0);
|
||||
xout1 = _mm_xor_si128(_mm_load_si128(input + 1), xout1);
|
||||
xout2 = _mm_xor_si128(_mm_load_si128(input + 2), xout2);
|
||||
xout3 = _mm_xor_si128(_mm_load_si128(input + 3), xout3);
|
||||
|
||||
constexpr int input_increment = (64 << interleave) / sizeof(__m128i);
|
||||
|
||||
xout4 = _mm_xor_si128(_mm_load_si128(input + input_increment + 0), xout4);
|
||||
xout5 = _mm_xor_si128(_mm_load_si128(input + input_increment + 1), xout5);
|
||||
xout6 = _mm_xor_si128(_mm_load_si128(input + input_increment + 2), xout6);
|
||||
xout7 = _mm_xor_si128(_mm_load_si128(input + input_increment + 3), xout7);
|
||||
|
||||
input += input_increment * 2;
|
||||
i += 8;
|
||||
|
||||
if ((interleave > 0) && (i < props.memory() / sizeof(__m128i))) {
|
||||
_mm_prefetch((const char*)(input), _MM_HINT_T0);
|
||||
_mm_prefetch((const char*)(input + (64 << interleave) / sizeof(__m128i)), _MM_HINT_T0);
|
||||
for (size_t part = 0; part < (props.half_mem() ? 2 : 1); ++part) {
|
||||
if (props.half_mem() && (part == 1)) {
|
||||
input = input_begin;
|
||||
ctx->first_half = false;
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, interleave>(ctx);
|
||||
}
|
||||
|
||||
aes_round<SOFT_AES>(k0, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k1, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k2, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k3, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k4, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k5, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k6, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k7, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k8, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k9, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
for (size_t i = 0; i < N;) {
|
||||
xout0 = _mm_xor_si128(_mm_load_si128(input + 0), xout0);
|
||||
xout1 = _mm_xor_si128(_mm_load_si128(input + 1), xout1);
|
||||
xout2 = _mm_xor_si128(_mm_load_si128(input + 2), xout2);
|
||||
xout3 = _mm_xor_si128(_mm_load_si128(input + 3), xout3);
|
||||
|
||||
if (IS_HEAVY) {
|
||||
mix_and_propagate(xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7);
|
||||
constexpr int input_increment = (64 << interleave) / sizeof(__m128i);
|
||||
|
||||
xout4 = _mm_xor_si128(_mm_load_si128(input + input_increment + 0), xout4);
|
||||
xout5 = _mm_xor_si128(_mm_load_si128(input + input_increment + 1), xout5);
|
||||
xout6 = _mm_xor_si128(_mm_load_si128(input + input_increment + 2), xout6);
|
||||
xout7 = _mm_xor_si128(_mm_load_si128(input + input_increment + 3), xout7);
|
||||
|
||||
input += input_increment * 2;
|
||||
i += 8;
|
||||
|
||||
if (i < N) {
|
||||
_mm_prefetch((const char*)(input), _MM_HINT_T0);
|
||||
_mm_prefetch((const char*)(input + input_increment), _MM_HINT_T0);
|
||||
}
|
||||
|
||||
aes_round<SOFT_AES>(k0, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k1, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k2, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k3, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k4, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k5, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k6, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k7, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k8, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
aes_round<SOFT_AES>(k9, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
|
||||
|
||||
if (IS_HEAVY) {
|
||||
mix_and_propagate(xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if (IS_HEAVY) {
|
||||
input = input_begin;
|
||||
for (size_t i = 0; i < props.memory() / sizeof(__m128i);) {
|
||||
for (size_t i = 0; i < N;) {
|
||||
xout0 = _mm_xor_si128(_mm_load_si128(input + 0), xout0);
|
||||
xout1 = _mm_xor_si128(_mm_load_si128(input + 1), xout1);
|
||||
xout2 = _mm_xor_si128(_mm_load_si128(input + 2), xout2);
|
||||
@@ -529,6 +579,9 @@ static inline __m128i int_sqrt_v2(const uint64_t n0)
|
||||
void v4_soft_aes_compile_code(const V4_Instruction *code, int code_size, void *machine_code, xmrig::Assembly ASM);
|
||||
|
||||
|
||||
alignas(64) static const uint32_t tweak1_table[256] = { 268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,268435456,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,805306368,0,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456,805306368,268435456 };
|
||||
|
||||
|
||||
namespace xmrig {
|
||||
|
||||
|
||||
@@ -547,12 +600,7 @@ static inline void cryptonight_monero_tweak(uint64_t *mem_out, const uint8_t *l,
|
||||
tmp = _mm_castps_si128(_mm_movehl_ps(_mm_castsi128_ps(tmp), _mm_castsi128_ps(tmp)));
|
||||
uint64_t vh = _mm_cvtsi128_si64(tmp);
|
||||
|
||||
uint8_t x = static_cast<uint8_t>(vh >> 24);
|
||||
static const uint16_t table = 0x7531;
|
||||
const uint8_t index = (((x >> (3)) & 6) | (x & 1)) << 1;
|
||||
vh ^= ((table >> index) & 0x3) << 28;
|
||||
|
||||
mem_out[1] = vh;
|
||||
mem_out[1] = vh ^ tweak1_table[static_cast<uint32_t>(vh) >> 24];
|
||||
}
|
||||
}
|
||||
|
||||
@@ -593,7 +641,11 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
}
|
||||
|
||||
keccak(input, size, ctx[0]->state);
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, interleave>(reinterpret_cast<const __m128i *>(ctx[0]->state), reinterpret_cast<__m128i *>(ctx[0]->memory));
|
||||
|
||||
if (props.half_mem()) {
|
||||
ctx[0]->first_half = true;
|
||||
}
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, interleave>(ctx[0]);
|
||||
|
||||
uint64_t *h0 = reinterpret_cast<uint64_t*>(ctx[0]->state);
|
||||
uint8_t *l0 = ctx[0]->memory;
|
||||
@@ -748,7 +800,7 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
}
|
||||
# endif
|
||||
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, interleave>(reinterpret_cast<const __m128i *>(ctx[0]->memory), reinterpret_cast<__m128i *>(ctx[0]->state));
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, interleave>(ctx[0]);
|
||||
keccakf(h0, 24);
|
||||
extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
|
||||
}
|
||||
@@ -906,7 +958,11 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
}
|
||||
|
||||
keccak(input, size, ctx[0]->state);
|
||||
cn_explode_scratchpad<ALGO, false, 0>(reinterpret_cast<const __m128i*>(ctx[0]->state), reinterpret_cast<__m128i*>(ctx[0]->memory));
|
||||
|
||||
if (props.half_mem()) {
|
||||
ctx[0]->first_half = true;
|
||||
}
|
||||
cn_explode_scratchpad<ALGO, false, 0>(ctx[0]);
|
||||
|
||||
if (ALGO == Algorithm::CN_2) {
|
||||
if (ASM == Assembly::INTEL) {
|
||||
@@ -988,7 +1044,7 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
ctx[0]->generated_code(ctx);
|
||||
}
|
||||
|
||||
cn_implode_scratchpad<ALGO, false, 0>(reinterpret_cast<const __m128i*>(ctx[0]->memory), reinterpret_cast<__m128i*>(ctx[0]->state));
|
||||
cn_implode_scratchpad<ALGO, false, 0>(ctx[0]);
|
||||
keccakf(reinterpret_cast<uint64_t*>(ctx[0]->state), 24);
|
||||
extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
|
||||
}
|
||||
@@ -1010,8 +1066,12 @@ inline void cryptonight_double_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
keccak(input, size, ctx[0]->state);
|
||||
keccak(input + size, size, ctx[1]->state);
|
||||
|
||||
cn_explode_scratchpad<ALGO, false, 0>(reinterpret_cast<const __m128i*>(ctx[0]->state), reinterpret_cast<__m128i*>(ctx[0]->memory));
|
||||
cn_explode_scratchpad<ALGO, false, 0>(reinterpret_cast<const __m128i*>(ctx[1]->state), reinterpret_cast<__m128i*>(ctx[1]->memory));
|
||||
if (props.half_mem()) {
|
||||
ctx[0]->first_half = true;
|
||||
ctx[1]->first_half = true;
|
||||
}
|
||||
cn_explode_scratchpad<ALGO, false, 0>(ctx[0]);
|
||||
cn_explode_scratchpad<ALGO, false, 0>(ctx[1]);
|
||||
|
||||
if (ALGO == Algorithm::CN_2) {
|
||||
cnv2_double_mainloop_sandybridge_asm(ctx);
|
||||
@@ -1050,8 +1110,8 @@ inline void cryptonight_double_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
ctx[0]->generated_code(ctx);
|
||||
}
|
||||
|
||||
cn_implode_scratchpad<ALGO, false, 0>(reinterpret_cast<const __m128i*>(ctx[0]->memory), reinterpret_cast<__m128i*>(ctx[0]->state));
|
||||
cn_implode_scratchpad<ALGO, false, 0>(reinterpret_cast<const __m128i*>(ctx[1]->memory), reinterpret_cast<__m128i*>(ctx[1]->state));
|
||||
cn_implode_scratchpad<ALGO, false, 0>(ctx[0]);
|
||||
cn_implode_scratchpad<ALGO, false, 0>(ctx[1]);
|
||||
|
||||
keccakf(reinterpret_cast<uint64_t*>(ctx[0]->state), 24);
|
||||
keccakf(reinterpret_cast<uint64_t*>(ctx[1]->state), 24);
|
||||
@@ -1102,8 +1162,12 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
VARIANT4_RANDOM_MATH_INIT(0);
|
||||
VARIANT4_RANDOM_MATH_INIT(1);
|
||||
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(reinterpret_cast<const __m128i *>(h0), reinterpret_cast<__m128i *>(l0));
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(reinterpret_cast<const __m128i *>(h1), reinterpret_cast<__m128i *>(l1));
|
||||
if (props.half_mem()) {
|
||||
ctx[0]->first_half = true;
|
||||
ctx[1]->first_half = true;
|
||||
}
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(ctx[0]);
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(ctx[1]);
|
||||
|
||||
uint64_t al0 = h0[0] ^ h0[4];
|
||||
uint64_t al1 = h1[0] ^ h1[4];
|
||||
@@ -1298,8 +1362,8 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
bx10 = cx1;
|
||||
}
|
||||
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(reinterpret_cast<const __m128i *>(l0), reinterpret_cast<__m128i *>(h0));
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(reinterpret_cast<const __m128i *>(l1), reinterpret_cast<__m128i *>(h1));
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(ctx[0]);
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(ctx[1]);
|
||||
|
||||
keccakf(h0, 24);
|
||||
keccakf(h1, 24);
|
||||
@@ -1309,6 +1373,198 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
}
|
||||
|
||||
|
||||
static inline void cryptonight_monero_tweak_gr(uint64_t* mem_out, const uint8_t* l, uint64_t idx, __m128i ax0, __m128i bx0, __m128i cx)
|
||||
{
|
||||
__m128i tmp = _mm_xor_si128(bx0, cx);
|
||||
mem_out[0] = _mm_cvtsi128_si64(tmp);
|
||||
|
||||
tmp = _mm_castps_si128(_mm_movehl_ps(_mm_castsi128_ps(tmp), _mm_castsi128_ps(tmp)));
|
||||
uint64_t vh = _mm_cvtsi128_si64(tmp);
|
||||
|
||||
mem_out[1] = vh ^ tweak1_table[static_cast<uint32_t>(vh) >> 24];
|
||||
}
|
||||
|
||||
|
||||
template<Algorithm::Id ALGO, bool SOFT_AES>
|
||||
void cryptonight_quad_hash_zen(const uint8_t* __restrict__ input, size_t size, uint8_t* __restrict__ output, cryptonight_ctx** __restrict__ ctx, uint64_t height)
|
||||
{
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
constexpr size_t MASK = props.mask();
|
||||
constexpr Algorithm::Id BASE = props.base();
|
||||
|
||||
if (BASE == Algorithm::CN_1 && size < 43) {
|
||||
memset(output, 0, 64);
|
||||
return;
|
||||
}
|
||||
|
||||
keccak(input + size * 0, size, ctx[0]->state);
|
||||
keccak(input + size * 1, size, ctx[1]->state);
|
||||
keccak(input + size * 2, size, ctx[2]->state);
|
||||
keccak(input + size * 3, size, ctx[3]->state);
|
||||
|
||||
uint8_t* l0 = ctx[0]->memory;
|
||||
uint8_t* l1 = ctx[1]->memory;
|
||||
uint8_t* l2 = ctx[2]->memory;
|
||||
uint8_t* l3 = ctx[3]->memory;
|
||||
|
||||
uint64_t* h0 = reinterpret_cast<uint64_t*>(ctx[0]->state);
|
||||
uint64_t* h1 = reinterpret_cast<uint64_t*>(ctx[1]->state);
|
||||
uint64_t* h2 = reinterpret_cast<uint64_t*>(ctx[2]->state);
|
||||
uint64_t* h3 = reinterpret_cast<uint64_t*>(ctx[3]->state);
|
||||
|
||||
VARIANT1_INIT(0);
|
||||
VARIANT1_INIT(1);
|
||||
VARIANT1_INIT(2);
|
||||
VARIANT1_INIT(3);
|
||||
|
||||
if (props.half_mem()) {
|
||||
ctx[0]->first_half = true;
|
||||
ctx[1]->first_half = true;
|
||||
ctx[2]->first_half = true;
|
||||
ctx[3]->first_half = true;
|
||||
}
|
||||
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(ctx[0]);
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(ctx[1]);
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(ctx[2]);
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(ctx[3]);
|
||||
|
||||
uint64_t al0 = h0[0] ^ h0[4];
|
||||
uint64_t al1 = h1[0] ^ h1[4];
|
||||
uint64_t al2 = h2[0] ^ h2[4];
|
||||
uint64_t al3 = h3[0] ^ h3[4];
|
||||
|
||||
uint64_t ah0 = h0[1] ^ h0[5];
|
||||
uint64_t ah1 = h1[1] ^ h1[5];
|
||||
uint64_t ah2 = h2[1] ^ h2[5];
|
||||
uint64_t ah3 = h3[1] ^ h3[5];
|
||||
|
||||
__m128i bx00 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
|
||||
__m128i bx10 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
|
||||
__m128i bx20 = _mm_set_epi64x(h2[3] ^ h2[7], h2[2] ^ h2[6]);
|
||||
__m128i bx30 = _mm_set_epi64x(h3[3] ^ h3[7], h3[2] ^ h3[6]);
|
||||
|
||||
uint64_t idx0 = al0;
|
||||
uint64_t idx1 = al1;
|
||||
uint64_t idx2 = al2;
|
||||
uint64_t idx3 = al3;
|
||||
|
||||
__m128i cx0, cx1, cx2, cx3;
|
||||
|
||||
if (!SOFT_AES) {
|
||||
cx0 = _mm_load_si128(reinterpret_cast<const __m128i*>(&l0[idx0 & MASK]));
|
||||
cx1 = _mm_load_si128(reinterpret_cast<const __m128i*>(&l1[idx1 & MASK]));
|
||||
cx2 = _mm_load_si128(reinterpret_cast<const __m128i*>(&l2[idx2 & MASK]));
|
||||
cx3 = _mm_load_si128(reinterpret_cast<const __m128i*>(&l3[idx3 & MASK]));
|
||||
}
|
||||
|
||||
for (size_t i = 0; i < props.iterations(); i++) {
|
||||
const __m128i ax0 = _mm_set_epi64x(ah0, al0);
|
||||
const __m128i ax1 = _mm_set_epi64x(ah1, al1);
|
||||
const __m128i ax2 = _mm_set_epi64x(ah2, al2);
|
||||
const __m128i ax3 = _mm_set_epi64x(ah3, al3);
|
||||
|
||||
if (SOFT_AES) {
|
||||
cx0 = soft_aesenc(&l0[idx0 & MASK], ax0, reinterpret_cast<const uint32_t*>(saes_table));
|
||||
cx1 = soft_aesenc(&l1[idx1 & MASK], ax1, reinterpret_cast<const uint32_t*>(saes_table));
|
||||
cx2 = soft_aesenc(&l2[idx2 & MASK], ax2, reinterpret_cast<const uint32_t*>(saes_table));
|
||||
cx3 = soft_aesenc(&l3[idx3 & MASK], ax3, reinterpret_cast<const uint32_t*>(saes_table));
|
||||
}
|
||||
else {
|
||||
cx0 = _mm_aesenc_si128(cx0, ax0);
|
||||
cx1 = _mm_aesenc_si128(cx1, ax1);
|
||||
cx2 = _mm_aesenc_si128(cx2, ax2);
|
||||
cx3 = _mm_aesenc_si128(cx3, ax3);
|
||||
if (MASK > 131072) {
|
||||
_mm_prefetch((const char*)(&l0[_mm_cvtsi128_si32(cx0) & MASK]), _MM_HINT_T0);
|
||||
_mm_prefetch((const char*)(&l1[_mm_cvtsi128_si32(cx1) & MASK]), _MM_HINT_T0);
|
||||
_mm_prefetch((const char*)(&l2[_mm_cvtsi128_si32(cx2) & MASK]), _MM_HINT_T0);
|
||||
_mm_prefetch((const char*)(&l3[_mm_cvtsi128_si32(cx3) & MASK]), _MM_HINT_T0);
|
||||
}
|
||||
}
|
||||
|
||||
cryptonight_monero_tweak_gr((uint64_t*)&l0[idx0 & MASK], l0, idx0 & MASK, ax0, bx00, cx0);
|
||||
cryptonight_monero_tweak_gr((uint64_t*)&l1[idx1 & MASK], l1, idx1 & MASK, ax1, bx10, cx1);
|
||||
cryptonight_monero_tweak_gr((uint64_t*)&l2[idx2 & MASK], l2, idx2 & MASK, ax2, bx20, cx2);
|
||||
cryptonight_monero_tweak_gr((uint64_t*)&l3[idx3 & MASK], l3, idx3 & MASK, ax3, bx30, cx3);
|
||||
|
||||
idx0 = _mm_cvtsi128_si64(cx0);
|
||||
idx1 = _mm_cvtsi128_si64(cx1);
|
||||
idx2 = _mm_cvtsi128_si64(cx2);
|
||||
idx3 = _mm_cvtsi128_si64(cx3);
|
||||
|
||||
uint64_t hi, lo, cl, ch;
|
||||
|
||||
cl = ((uint64_t*)&l0[idx0 & MASK])[0];
|
||||
ch = ((uint64_t*)&l0[idx0 & MASK])[1];
|
||||
lo = __umul128(idx0, cl, &hi);
|
||||
al0 += hi;
|
||||
ah0 += lo;
|
||||
((uint64_t*)&l0[idx0 & MASK])[0] = al0;
|
||||
((uint64_t*)&l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0;
|
||||
al0 ^= cl;
|
||||
ah0 ^= ch;
|
||||
idx0 = al0;
|
||||
bx00 = cx0;
|
||||
if (!SOFT_AES) cx0 = _mm_load_si128(reinterpret_cast<const __m128i*>(&l0[idx0 & MASK]));
|
||||
|
||||
cl = ((uint64_t*)&l1[idx1 & MASK])[0];
|
||||
ch = ((uint64_t*)&l1[idx1 & MASK])[1];
|
||||
lo = __umul128(idx1, cl, &hi);
|
||||
al1 += hi;
|
||||
ah1 += lo;
|
||||
((uint64_t*)&l1[idx1 & MASK])[0] = al1;
|
||||
((uint64_t*)&l1[idx1 & MASK])[1] = ah1 ^ tweak1_2_1;
|
||||
al1 ^= cl;
|
||||
ah1 ^= ch;
|
||||
idx1 = al1;
|
||||
bx10 = cx1;
|
||||
if (!SOFT_AES) cx1 = _mm_load_si128(reinterpret_cast<const __m128i*>(&l1[idx1 & MASK]));
|
||||
|
||||
cl = ((uint64_t*)&l2[idx2 & MASK])[0];
|
||||
ch = ((uint64_t*)&l2[idx2 & MASK])[1];
|
||||
lo = __umul128(idx2, cl, &hi);
|
||||
al2 += hi;
|
||||
ah2 += lo;
|
||||
((uint64_t*)&l2[idx2 & MASK])[0] = al2;
|
||||
((uint64_t*)&l2[idx2 & MASK])[1] = ah2 ^ tweak1_2_2;
|
||||
al2 ^= cl;
|
||||
ah2 ^= ch;
|
||||
idx2 = al2;
|
||||
bx20 = cx2;
|
||||
if (!SOFT_AES) cx2 = _mm_load_si128(reinterpret_cast<const __m128i*>(&l2[idx2 & MASK]));
|
||||
|
||||
cl = ((uint64_t*)&l3[idx3 & MASK])[0];
|
||||
ch = ((uint64_t*)&l3[idx3 & MASK])[1];
|
||||
lo = __umul128(idx3, cl, &hi);
|
||||
al3 += hi;
|
||||
ah3 += lo;
|
||||
((uint64_t*)&l3[idx3 & MASK])[0] = al3;
|
||||
((uint64_t*)&l3[idx3 & MASK])[1] = ah3 ^ tweak1_2_3;
|
||||
al3 ^= cl;
|
||||
ah3 ^= ch;
|
||||
idx3 = al3;
|
||||
bx30 = cx3;
|
||||
if (!SOFT_AES) cx3 = _mm_load_si128(reinterpret_cast<const __m128i*>(&l3[idx3 & MASK]));
|
||||
}
|
||||
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(ctx[0]);
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(ctx[1]);
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(ctx[2]);
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(ctx[3]);
|
||||
|
||||
keccakf(h0, 24);
|
||||
keccakf(h1, 24);
|
||||
keccakf(h2, 24);
|
||||
keccakf(h3, 24);
|
||||
|
||||
extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
|
||||
extra_hashes[ctx[1]->state[0] & 3](ctx[1]->state, 200, output + 32);
|
||||
extra_hashes[ctx[2]->state[0] & 3](ctx[2]->state, 200, output + 64);
|
||||
extra_hashes[ctx[3]->state[0] & 3](ctx[3]->state, 200, output + 96);
|
||||
}
|
||||
|
||||
|
||||
#define CN_STEP1(a, b0, b1, c, l, ptr, idx, conc_var) \
|
||||
ptr = reinterpret_cast<__m128i*>(&l[idx & MASK]); \
|
||||
c = _mm_load_si128(ptr); \
|
||||
@@ -1444,7 +1700,10 @@ inline void cryptonight_triple_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
for (size_t i = 0; i < 3; i++) {
|
||||
keccak(input + size * i, size, ctx[i]->state);
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(reinterpret_cast<const __m128i*>(ctx[i]->state), reinterpret_cast<__m128i*>(ctx[i]->memory));
|
||||
if (props.half_mem()) {
|
||||
ctx[i]->first_half = true;
|
||||
}
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(ctx[i]);
|
||||
}
|
||||
|
||||
uint8_t* l0 = ctx[0]->memory;
|
||||
@@ -1489,7 +1748,7 @@ inline void cryptonight_triple_hash(const uint8_t *__restrict__ input, size_t si
|
||||
}
|
||||
|
||||
for (size_t i = 0; i < 3; i++) {
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(reinterpret_cast<const __m128i*>(ctx[i]->memory), reinterpret_cast<__m128i*>(ctx[i]->state));
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(ctx[i]);
|
||||
keccakf(reinterpret_cast<uint64_t*>(ctx[i]->state), 24);
|
||||
extra_hashes[ctx[i]->state[0] & 3](ctx[i]->state, 200, output + 32 * i);
|
||||
}
|
||||
@@ -1499,6 +1758,14 @@ inline void cryptonight_triple_hash(const uint8_t *__restrict__ input, size_t si
|
||||
template<Algorithm::Id ALGO, bool SOFT_AES>
|
||||
inline void cryptonight_quad_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height)
|
||||
{
|
||||
const auto arch = Cpu::info()->arch();
|
||||
if ((arch >= ICpuInfo::ARCH_ZEN) && (arch <= ICpuInfo::ARCH_ZEN3)) {
|
||||
if ((ALGO == Algorithm::CN_GR_0) || (ALGO == Algorithm::CN_GR_1) || (ALGO == Algorithm::CN_GR_2) || (ALGO == Algorithm::CN_GR_3) || (ALGO == Algorithm::CN_GR_4) || (ALGO == Algorithm::CN_GR_5)) {
|
||||
cryptonight_quad_hash_zen<ALGO, SOFT_AES>(input, size, output, ctx, height);
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
constexpr size_t MASK = props.mask();
|
||||
constexpr Algorithm::Id BASE = props.base();
|
||||
@@ -1518,7 +1785,10 @@ inline void cryptonight_quad_hash(const uint8_t *__restrict__ input, size_t size
|
||||
|
||||
for (size_t i = 0; i < 4; i++) {
|
||||
keccak(input + size * i, size, ctx[i]->state);
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(reinterpret_cast<const __m128i*>(ctx[i]->state), reinterpret_cast<__m128i*>(ctx[i]->memory));
|
||||
if (props.half_mem()) {
|
||||
ctx[i]->first_half = true;
|
||||
}
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(ctx[i]);
|
||||
}
|
||||
|
||||
uint8_t* l0 = ctx[0]->memory;
|
||||
@@ -1571,7 +1841,7 @@ inline void cryptonight_quad_hash(const uint8_t *__restrict__ input, size_t size
|
||||
}
|
||||
|
||||
for (size_t i = 0; i < 4; i++) {
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(reinterpret_cast<const __m128i*>(ctx[i]->memory), reinterpret_cast<__m128i*>(ctx[i]->state));
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(ctx[i]);
|
||||
keccakf(reinterpret_cast<uint64_t*>(ctx[i]->state), 24);
|
||||
extra_hashes[ctx[i]->state[0] & 3](ctx[i]->state, 200, output + 32 * i);
|
||||
}
|
||||
@@ -1600,7 +1870,10 @@ inline void cryptonight_penta_hash(const uint8_t *__restrict__ input, size_t siz
|
||||
|
||||
for (size_t i = 0; i < 5; i++) {
|
||||
keccak(input + size * i, size, ctx[i]->state);
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(reinterpret_cast<const __m128i*>(ctx[i]->state), reinterpret_cast<__m128i*>(ctx[i]->memory));
|
||||
if (props.half_mem()) {
|
||||
ctx[i]->first_half = true;
|
||||
}
|
||||
cn_explode_scratchpad<ALGO, SOFT_AES, 0>(ctx[i]);
|
||||
}
|
||||
|
||||
uint8_t* l0 = ctx[0]->memory;
|
||||
@@ -1661,7 +1934,7 @@ inline void cryptonight_penta_hash(const uint8_t *__restrict__ input, size_t siz
|
||||
}
|
||||
|
||||
for (size_t i = 0; i < 5; i++) {
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(reinterpret_cast<const __m128i*>(ctx[i]->memory), reinterpret_cast<__m128i*>(ctx[i]->state));
|
||||
cn_implode_scratchpad<ALGO, SOFT_AES, 0>(ctx[i]);
|
||||
keccakf(reinterpret_cast<uint64_t*>(ctx[i]->state), 24);
|
||||
extra_hashes[ctx[i]->state[0] & 3](ctx[i]->state, 200, output + 32 * i);
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user