mirror of
https://github.com/xmrig/xmrig.git
synced 2025-12-10 09:02:45 -05:00
Compare commits
5 Commits
a016146e4e
...
639aa0ca15
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
639aa0ca15 | ||
|
|
77e2f3a028 | ||
|
|
206295c6cb | ||
|
|
862280f28c | ||
|
|
814e1de2a6 |
@@ -138,6 +138,93 @@ __kernel void blake2b_initial_hash(__global void *out, __global const void* bloc
|
||||
t[7] = hash[7];
|
||||
}
|
||||
|
||||
void blake2b_512_process_double_block_variable(ulong *out, ulong* m, __global const ulong* in, uint in_len, uint out_len)
|
||||
{
|
||||
ulong v[16] =
|
||||
{
|
||||
iv0 ^ (0x01010000u | out_len), iv1, iv2, iv3, iv4 , iv5, iv6, iv7,
|
||||
iv0 , iv1, iv2, iv3, iv4 ^ 128, iv5, iv6, iv7,
|
||||
};
|
||||
|
||||
BLAKE2B_ROUNDS();
|
||||
|
||||
ulong h[8];
|
||||
v[0] = h[0] = v[0] ^ v[8] ^ iv0 ^ (0x01010000u | out_len);
|
||||
v[1] = h[1] = v[1] ^ v[9] ^ iv1;
|
||||
v[2] = h[2] = v[2] ^ v[10] ^ iv2;
|
||||
v[3] = h[3] = v[3] ^ v[11] ^ iv3;
|
||||
v[4] = h[4] = v[4] ^ v[12] ^ iv4;
|
||||
v[5] = h[5] = v[5] ^ v[13] ^ iv5;
|
||||
v[6] = h[6] = v[6] ^ v[14] ^ iv6;
|
||||
v[7] = h[7] = v[7] ^ v[15] ^ iv7;
|
||||
v[8] = iv0;
|
||||
v[9] = iv1;
|
||||
v[10] = iv2;
|
||||
v[11] = iv3;
|
||||
v[12] = iv4 ^ in_len;
|
||||
v[13] = iv5;
|
||||
v[14] = ~iv6;
|
||||
v[15] = iv7;
|
||||
|
||||
m[ 0] = (in_len > 128) ? in[16] : 0;
|
||||
m[ 1] = (in_len > 136) ? in[17] : 0;
|
||||
m[ 2] = (in_len > 144) ? in[18] : 0;
|
||||
m[ 3] = (in_len > 152) ? in[19] : 0;
|
||||
m[ 4] = (in_len > 160) ? in[20] : 0;
|
||||
m[ 5] = (in_len > 168) ? in[21] : 0;
|
||||
m[ 6] = (in_len > 176) ? in[22] : 0;
|
||||
m[ 7] = (in_len > 184) ? in[23] : 0;
|
||||
m[ 8] = (in_len > 192) ? in[24] : 0;
|
||||
m[ 9] = (in_len > 200) ? in[25] : 0;
|
||||
m[10] = (in_len > 208) ? in[26] : 0;
|
||||
m[11] = (in_len > 216) ? in[27] : 0;
|
||||
m[12] = (in_len > 224) ? in[28] : 0;
|
||||
m[13] = (in_len > 232) ? in[29] : 0;
|
||||
m[14] = (in_len > 240) ? in[30] : 0;
|
||||
m[15] = (in_len > 248) ? in[31] : 0;
|
||||
|
||||
if (in_len % sizeof(ulong))
|
||||
m[(in_len - 128) / sizeof(ulong)] &= (ulong)(-1) >> (64 - (in_len % sizeof(ulong)) * 8);
|
||||
|
||||
BLAKE2B_ROUNDS();
|
||||
|
||||
if (out_len > 0) out[0] = h[0] ^ v[0] ^ v[8];
|
||||
if (out_len > 8) out[1] = h[1] ^ v[1] ^ v[9];
|
||||
if (out_len > 16) out[2] = h[2] ^ v[2] ^ v[10];
|
||||
if (out_len > 24) out[3] = h[3] ^ v[3] ^ v[11];
|
||||
if (out_len > 32) out[4] = h[4] ^ v[4] ^ v[12];
|
||||
if (out_len > 40) out[5] = h[5] ^ v[5] ^ v[13];
|
||||
if (out_len > 48) out[6] = h[6] ^ v[6] ^ v[14];
|
||||
if (out_len > 56) out[7] = h[7] ^ v[7] ^ v[15];
|
||||
}
|
||||
|
||||
__attribute__((reqd_work_group_size(64, 1, 1)))
|
||||
__kernel void blake2b_initial_hash_double(__global void *out, __global const void* blockTemplate, uint blockTemplateSize, uint start_nonce)
|
||||
{
|
||||
const uint global_index = get_global_id(0);
|
||||
|
||||
__global const ulong* p = (__global const ulong*) blockTemplate;
|
||||
|
||||
ulong m[16] = { p[0], p[1], p[2], p[3], p[4], p[5], p[6], p[7], p[8], p[9], p[10], p[11], p[12], p[13], p[14], p[15] };
|
||||
|
||||
const ulong nonce = start_nonce + global_index;
|
||||
m[4] = (m[4] & ((ulong)(-1) >> 8)) | (nonce << 56);
|
||||
m[5] = (m[5] & ((ulong)(-1) << 24)) | (nonce >> 8);
|
||||
|
||||
ulong hash[8];
|
||||
blake2b_512_process_double_block_variable(hash, m, p, blockTemplateSize, 64);
|
||||
|
||||
__global ulong* t = ((__global ulong*) out) + global_index * 8;
|
||||
t[0] = hash[0];
|
||||
t[1] = hash[1];
|
||||
t[2] = hash[2];
|
||||
t[3] = hash[3];
|
||||
t[4] = hash[4];
|
||||
t[5] = hash[5];
|
||||
t[6] = hash[6];
|
||||
t[7] = hash[7];
|
||||
}
|
||||
|
||||
#define in_len 256
|
||||
|
||||
#define out_len 32
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,58 @@
|
||||
/* XMRig
|
||||
* Copyright 2010 Jeff Garzik <jgarzik@pobox.com>
|
||||
* Copyright 2012-2014 pooler <pooler@litecoinpool.org>
|
||||
* Copyright 2014 Lucas Jones <https://github.com/lucasjones>
|
||||
* Copyright 2014-2016 Wolf9466 <https://github.com/OhGodAPet>
|
||||
* Copyright 2016 Jay D Dee <jayddee246@gmail.com>
|
||||
* Copyright 2017-2018 XMR-Stak <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
|
||||
* Copyright 2018-2019 SChernykh <https://github.com/SChernykh>
|
||||
* Copyright 2016-2019 XMRig <https://github.com/xmrig>, <support@xmrig.com>
|
||||
*
|
||||
* This program is free software: you can redistribute it and/or modify
|
||||
* it under the terms of the GNU General Public License as published by
|
||||
* the Free Software Foundation, either version 3 of the License, or
|
||||
* (at your option) any later version.
|
||||
*
|
||||
* This program is distributed in the hope that it will be useful,
|
||||
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
||||
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
||||
* GNU General Public License for more details.
|
||||
*
|
||||
* You should have received a copy of the GNU General Public License
|
||||
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
||||
*/
|
||||
|
||||
|
||||
#include "backend/opencl/kernels/rx/Blake2bInitialHashDoubleKernel.h"
|
||||
#include "backend/opencl/wrappers/OclLib.h"
|
||||
|
||||
|
||||
void xmrig::Blake2bInitialHashDoubleKernel::enqueue(cl_command_queue queue, size_t threads)
|
||||
{
|
||||
const size_t gthreads = threads;
|
||||
static const size_t lthreads = 64;
|
||||
|
||||
enqueueNDRange(queue, 1, nullptr, >hreads, <hreads);
|
||||
}
|
||||
|
||||
|
||||
// __kernel void blake2b_initial_hash_double(__global void *out, __global const void* blockTemplate, uint blockTemplateSize, uint start_nonce)
|
||||
void xmrig::Blake2bInitialHashDoubleKernel::setArgs(cl_mem out, cl_mem blockTemplate)
|
||||
{
|
||||
setArg(0, sizeof(cl_mem), &out);
|
||||
setArg(1, sizeof(cl_mem), &blockTemplate);
|
||||
}
|
||||
|
||||
|
||||
void xmrig::Blake2bInitialHashDoubleKernel::setBlobSize(size_t size)
|
||||
{
|
||||
const uint32_t s = size;
|
||||
|
||||
setArg(2, sizeof(uint32_t), &s);
|
||||
}
|
||||
|
||||
|
||||
void xmrig::Blake2bInitialHashDoubleKernel::setNonce(uint32_t nonce)
|
||||
{
|
||||
setArg(3, sizeof(uint32_t), &nonce);
|
||||
}
|
||||
@@ -0,0 +1,50 @@
|
||||
/* XMRig
|
||||
* Copyright 2010 Jeff Garzik <jgarzik@pobox.com>
|
||||
* Copyright 2012-2014 pooler <pooler@litecoinpool.org>
|
||||
* Copyright 2014 Lucas Jones <https://github.com/lucasjones>
|
||||
* Copyright 2014-2016 Wolf9466 <https://github.com/OhGodAPet>
|
||||
* Copyright 2016 Jay D Dee <jayddee246@gmail.com>
|
||||
* Copyright 2017-2018 XMR-Stak <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
|
||||
* Copyright 2018-2019 SChernykh <https://github.com/SChernykh>
|
||||
* Copyright 2016-2019 XMRig <https://github.com/xmrig>, <support@xmrig.com>
|
||||
*
|
||||
* This program is free software: you can redistribute it and/or modify
|
||||
* it under the terms of the GNU General Public License as published by
|
||||
* the Free Software Foundation, either version 3 of the License, or
|
||||
* (at your option) any later version.
|
||||
*
|
||||
* This program is distributed in the hope that it will be useful,
|
||||
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
||||
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
||||
* GNU General Public License for more details.
|
||||
*
|
||||
* You should have received a copy of the GNU General Public License
|
||||
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
||||
*/
|
||||
|
||||
#ifndef XMRIG_BLAKE2BINITIALHASHDOUBLEKERNEL_H
|
||||
#define XMRIG_BLAKE2BINITIALHASHDOUBLEKERNEL_H
|
||||
|
||||
|
||||
#include "backend/opencl/wrappers/OclKernel.h"
|
||||
|
||||
|
||||
namespace xmrig {
|
||||
|
||||
|
||||
class Blake2bInitialHashDoubleKernel : public OclKernel
|
||||
{
|
||||
public:
|
||||
inline Blake2bInitialHashDoubleKernel(cl_program program) : OclKernel(program, "blake2b_initial_hash_double") {}
|
||||
|
||||
void enqueue(cl_command_queue queue, size_t threads);
|
||||
void setArgs(cl_mem out, cl_mem blockTemplate);
|
||||
void setBlobSize(size_t size);
|
||||
void setNonce(uint32_t nonce);
|
||||
};
|
||||
|
||||
|
||||
} // namespace xmrig
|
||||
|
||||
|
||||
#endif /* XMRIG_BLAKE2BINITIALHASHDOUBLEKERNEL_H */
|
||||
@@ -80,6 +80,7 @@ if (WITH_OPENCL)
|
||||
if (WITH_RANDOMX)
|
||||
list(APPEND HEADERS_BACKEND_OPENCL
|
||||
src/backend/opencl/kernels/rx/Blake2bHashRegistersKernel.h
|
||||
src/backend/opencl/kernels/rx/Blake2bInitialHashDoubleKernel.h
|
||||
src/backend/opencl/kernels/rx/Blake2bInitialHashKernel.h
|
||||
src/backend/opencl/kernels/rx/ExecuteVmKernel.h
|
||||
src/backend/opencl/kernels/rx/FillAesKernel.h
|
||||
@@ -96,6 +97,7 @@ if (WITH_OPENCL)
|
||||
list(APPEND SOURCES_BACKEND_OPENCL
|
||||
src/backend/opencl/generators/ocl_generic_rx_generator.cpp
|
||||
src/backend/opencl/kernels/rx/Blake2bHashRegistersKernel.cpp
|
||||
src/backend/opencl/kernels/rx/Blake2bInitialHashDoubleKernel.cpp
|
||||
src/backend/opencl/kernels/rx/Blake2bInitialHashKernel.cpp
|
||||
src/backend/opencl/kernels/rx/ExecuteVmKernel.cpp
|
||||
src/backend/opencl/kernels/rx/FillAesKernel.cpp
|
||||
|
||||
@@ -25,6 +25,7 @@
|
||||
#include "backend/opencl/runners/OclRxBaseRunner.h"
|
||||
#include "backend/opencl/kernels/rx/Blake2bHashRegistersKernel.h"
|
||||
#include "backend/opencl/kernels/rx/Blake2bInitialHashKernel.h"
|
||||
#include "backend/opencl/kernels/rx/Blake2bInitialHashDoubleKernel.h"
|
||||
#include "backend/opencl/kernels/rx/FillAesKernel.h"
|
||||
#include "backend/opencl/kernels/rx/FindSharesKernel.h"
|
||||
#include "backend/opencl/kernels/rx/HashAesKernel.h"
|
||||
@@ -71,6 +72,7 @@ xmrig::OclRxBaseRunner::~OclRxBaseRunner()
|
||||
delete m_fillAes4Rx4_entropy;
|
||||
delete m_hashAes1Rx4;
|
||||
delete m_blake2b_initial_hash;
|
||||
delete m_blake2b_initial_hash_double;
|
||||
delete m_blake2b_hash_registers_32;
|
||||
delete m_blake2b_hash_registers_64;
|
||||
delete m_find_shares;
|
||||
@@ -87,12 +89,28 @@ void xmrig::OclRxBaseRunner::run(uint32_t nonce, uint32_t *hashOutput)
|
||||
{
|
||||
static const uint32_t zero = 0;
|
||||
|
||||
m_blake2b_initial_hash->setNonce(nonce);
|
||||
if (m_jobSize <= 128) {
|
||||
m_blake2b_initial_hash->setNonce(nonce);
|
||||
}
|
||||
else if (m_jobSize <= 256) {
|
||||
m_blake2b_initial_hash_double->setNonce(nonce);
|
||||
}
|
||||
else {
|
||||
hashOutput[0xFF] = 0;
|
||||
return;
|
||||
}
|
||||
|
||||
m_find_shares->setNonce(nonce);
|
||||
|
||||
enqueueWriteBuffer(m_output, CL_FALSE, sizeof(cl_uint) * 0xFF, sizeof(uint32_t), &zero);
|
||||
|
||||
m_blake2b_initial_hash->enqueue(m_queue, m_intensity);
|
||||
if (m_jobSize <= 128) {
|
||||
m_blake2b_initial_hash->enqueue(m_queue, m_intensity);
|
||||
}
|
||||
else {
|
||||
m_blake2b_initial_hash_double->enqueue(m_queue, m_intensity);
|
||||
}
|
||||
|
||||
m_fillAes1Rx4_scratchpad->enqueue(m_queue, m_intensity);
|
||||
|
||||
const uint32_t programCount = RxAlgo::programCount(m_algorithm);
|
||||
@@ -134,7 +152,11 @@ void xmrig::OclRxBaseRunner::set(const Job &job, uint8_t *blob)
|
||||
|
||||
enqueueWriteBuffer(m_input, CL_TRUE, 0, Job::kMaxBlobSize, blob);
|
||||
|
||||
m_jobSize = job.size();
|
||||
|
||||
m_blake2b_initial_hash->setBlobSize(job.size());
|
||||
m_blake2b_initial_hash_double->setBlobSize(job.size());
|
||||
|
||||
m_find_shares->setTarget(job.target());
|
||||
}
|
||||
|
||||
@@ -166,6 +188,9 @@ void xmrig::OclRxBaseRunner::build()
|
||||
m_blake2b_initial_hash = new Blake2bInitialHashKernel(m_program);
|
||||
m_blake2b_initial_hash->setArgs(m_hashes, m_input);
|
||||
|
||||
m_blake2b_initial_hash_double = new Blake2bInitialHashDoubleKernel(m_program);
|
||||
m_blake2b_initial_hash_double->setArgs(m_hashes, m_input);
|
||||
|
||||
m_blake2b_hash_registers_32 = new Blake2bHashRegistersKernel(m_program, "blake2b_hash_registers_32");
|
||||
m_blake2b_hash_registers_64 = new Blake2bHashRegistersKernel(m_program, "blake2b_hash_registers_64");
|
||||
|
||||
|
||||
@@ -35,6 +35,7 @@ namespace xmrig {
|
||||
|
||||
class Blake2bHashRegistersKernel;
|
||||
class Blake2bInitialHashKernel;
|
||||
class Blake2bInitialHashDoubleKernel;
|
||||
class FillAesKernel;
|
||||
class FindSharesKernel;
|
||||
class HashAesKernel;
|
||||
@@ -58,21 +59,24 @@ protected:
|
||||
protected:
|
||||
virtual void execute(uint32_t iteration) = 0;
|
||||
|
||||
Blake2bHashRegistersKernel *m_blake2b_hash_registers_32 = nullptr;
|
||||
Blake2bHashRegistersKernel *m_blake2b_hash_registers_64 = nullptr;
|
||||
Blake2bInitialHashKernel *m_blake2b_initial_hash = nullptr;
|
||||
Blake2bHashRegistersKernel *m_blake2b_hash_registers_32 = nullptr;
|
||||
Blake2bHashRegistersKernel *m_blake2b_hash_registers_64 = nullptr;
|
||||
Blake2bInitialHashKernel *m_blake2b_initial_hash = nullptr;
|
||||
Blake2bInitialHashDoubleKernel *m_blake2b_initial_hash_double = nullptr;
|
||||
Buffer m_seed;
|
||||
cl_mem m_dataset = nullptr;
|
||||
cl_mem m_entropy = nullptr;
|
||||
cl_mem m_hashes = nullptr;
|
||||
cl_mem m_rounding = nullptr;
|
||||
cl_mem m_scratchpads = nullptr;
|
||||
FillAesKernel *m_fillAes1Rx4_scratchpad = nullptr;
|
||||
FillAesKernel *m_fillAes4Rx4_entropy = nullptr;
|
||||
FindSharesKernel *m_find_shares = nullptr;
|
||||
HashAesKernel *m_hashAes1Rx4 = nullptr;
|
||||
uint32_t m_gcn_version = 12;
|
||||
uint32_t m_worksize = 8;
|
||||
cl_mem m_dataset = nullptr;
|
||||
cl_mem m_entropy = nullptr;
|
||||
cl_mem m_hashes = nullptr;
|
||||
cl_mem m_rounding = nullptr;
|
||||
cl_mem m_scratchpads = nullptr;
|
||||
FillAesKernel *m_fillAes1Rx4_scratchpad = nullptr;
|
||||
FillAesKernel *m_fillAes4Rx4_entropy = nullptr;
|
||||
FindSharesKernel *m_find_shares = nullptr;
|
||||
HashAesKernel *m_hashAes1Rx4 = nullptr;
|
||||
uint32_t m_gcn_version = 12;
|
||||
uint32_t m_worksize = 8;
|
||||
|
||||
size_t m_jobSize = 0;
|
||||
};
|
||||
|
||||
|
||||
|
||||
@@ -37,14 +37,36 @@ class CnAlgo
|
||||
public:
|
||||
constexpr CnAlgo() {};
|
||||
|
||||
constexpr inline Algorithm::Id base() const { static_assert(Algorithm::isCN(ALGO), "invalid CRYPTONIGHT algorithm"); return Algorithm::base(ALGO); }
|
||||
constexpr inline bool isHeavy() const { return Algorithm::family(ALGO) == Algorithm::CN_HEAVY; }
|
||||
constexpr inline bool isR() const { return ALGO == Algorithm::CN_R; }
|
||||
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; }
|
||||
# define ASSERT_CN static_assert(Algorithm::isCN(ALGO), "invalid CRYPTONIGHT algorithm")
|
||||
constexpr inline Algorithm::Id base() const { ASSERT_CN; return Algorithm::base(ALGO); }
|
||||
constexpr inline size_t memory() const { ASSERT_CN; return Algorithm::l3(ALGO); }
|
||||
constexpr inline uint32_t iterations() const { ASSERT_CN; 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; }
|
||||
|
||||
constexpr inline bool isBase1() const { ASSERT_CN; return Algorithm::base(ALGO) == Algorithm::CN_1; }
|
||||
constexpr inline bool isBase2() const { ASSERT_CN; return Algorithm::base(ALGO) == Algorithm::CN_2; }
|
||||
constexpr inline bool is2() const { return ALGO == Algorithm::CN_2; }
|
||||
constexpr inline bool isR() const { return ALGO == Algorithm::CN_R; }
|
||||
constexpr inline bool isHalf() const { return ALGO == Algorithm::CN_HALF; }
|
||||
constexpr inline bool isRTO() const { return ALGO == Algorithm::CN_RTO; }
|
||||
constexpr inline bool isRWZ() const { return ALGO == Algorithm::CN_RWZ; }
|
||||
constexpr inline bool isZLS() const { return ALGO == Algorithm::CN_ZLS; }
|
||||
constexpr inline bool isDouble() const { return ALGO == Algorithm::CN_DOUBLE; }
|
||||
constexpr inline bool isCCX() const { return ALGO == Algorithm::CN_CCX; }
|
||||
constexpr inline bool isHeavy() const { ASSERT_CN; return Algorithm::family(ALGO) == Algorithm::CN_HEAVY; }
|
||||
constexpr inline bool isHeavyTube() const { return ALGO == Algorithm::CN_HEAVY_TUBE; }
|
||||
constexpr inline bool isHeavyXHV() const { return ALGO == Algorithm::CN_HEAVY_XHV; }
|
||||
constexpr inline bool isPico0() const { return ALGO == Algorithm::CN_PICO_0; }
|
||||
constexpr inline bool isPicoTLO() const { return ALGO == Algorithm::CN_PICO_TLO; }
|
||||
constexpr inline bool isUPX2() const { return ALGO == Algorithm::CN_UPX2; }
|
||||
constexpr inline bool isGR0() const { return ALGO == Algorithm::CN_GR_0; }
|
||||
constexpr inline bool isGR1() const { return ALGO == Algorithm::CN_GR_1; }
|
||||
constexpr inline bool isGR2() const { return ALGO == Algorithm::CN_GR_2; }
|
||||
constexpr inline bool isGR3() const { return ALGO == Algorithm::CN_GR_3; }
|
||||
constexpr inline bool isGR4() const { return ALGO == Algorithm::CN_GR_4; }
|
||||
constexpr inline bool isGR5() const { return ALGO == Algorithm::CN_GR_5; }
|
||||
|
||||
inline static uint32_t iterations(Algorithm::Id algo)
|
||||
{
|
||||
switch (algo) {
|
||||
|
||||
@@ -603,7 +603,7 @@ static inline void cryptonight_monero_tweak(uint64_t *mem_out, const uint8_t *l,
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
|
||||
if (props.base() == Algorithm::CN_2) {
|
||||
VARIANT2_SHUFFLE(l, idx, ax0, bx0, bx1, cx, (((ALGO == Algorithm::CN_RWZ) || (ALGO == Algorithm::CN_UPX2)) ? 1 : 0));
|
||||
VARIANT2_SHUFFLE(l, idx, ax0, bx0, bx1, cx, ((props.isRWZ() || props.isUPX2()) ? 1 : 0));
|
||||
_mm_store_si128(reinterpret_cast<__m128i *>(mem_out), _mm_xor_si128(bx0, cx));
|
||||
} else {
|
||||
__m128i tmp = _mm_xor_si128(bx0, cx);
|
||||
@@ -665,15 +665,8 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
constexpr size_t MASK = props.mask();
|
||||
constexpr Algorithm::Id BASE = props.base();
|
||||
|
||||
# ifdef XMRIG_ALGO_CN_HEAVY
|
||||
constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE;
|
||||
# else
|
||||
constexpr bool IS_CN_HEAVY_TUBE = false;
|
||||
# endif
|
||||
|
||||
if (BASE == Algorithm::CN_1 && size < 43) {
|
||||
if (props.isBase1() && size < 43) {
|
||||
memset(output, 0, 32);
|
||||
return;
|
||||
}
|
||||
@@ -694,10 +687,7 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
V4_Instruction code[256];
|
||||
const int code_size = v4_random_math_init<ALGO>(code, height);
|
||||
|
||||
if (ALGO == Algorithm::CN_R) {
|
||||
v4_soft_aes_compile_code(code, code_size, reinterpret_cast<void*>(ctx[0]->generated_code), Assembly::NONE);
|
||||
}
|
||||
|
||||
v4_soft_aes_compile_code(code, code_size, reinterpret_cast<void*>(ctx[0]->generated_code), Assembly::NONE);
|
||||
ctx[0]->generated_code_data = { ALGO, height };
|
||||
}
|
||||
|
||||
@@ -718,26 +708,26 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
__m128i bx1 = _mm_set_epi64x(static_cast<int64_t>(h0[9] ^ h0[11]), static_cast<int64_t>(h0[8] ^ h0[10]));
|
||||
|
||||
__m128 conc_var;
|
||||
if (ALGO == Algorithm::CN_CCX) {
|
||||
if (props.isCCX()) {
|
||||
conc_var = _mm_setzero_ps();
|
||||
RESTORE_ROUNDING_MODE();
|
||||
}
|
||||
|
||||
for (size_t i = 0; i < props.iterations(); i++) {
|
||||
__m128i cx;
|
||||
if (IS_CN_HEAVY_TUBE || !SOFT_AES) {
|
||||
if (props.isHeavyTube() || !SOFT_AES) {
|
||||
cx = _mm_load_si128(reinterpret_cast<const __m128i *>(&l0[interleaved_index<interleave>(idx0 & MASK)]));
|
||||
if (ALGO == Algorithm::CN_CCX) {
|
||||
if (props.isCCX()) {
|
||||
cryptonight_conceal_tweak(cx, conc_var);
|
||||
}
|
||||
}
|
||||
|
||||
const __m128i ax0 = _mm_set_epi64x(static_cast<int64_t>(ah0), static_cast<int64_t>(al0));
|
||||
if (IS_CN_HEAVY_TUBE) {
|
||||
if (props.isHeavyTube()) {
|
||||
cx = aes_round_tweak_div(cx, ax0);
|
||||
}
|
||||
else if (SOFT_AES) {
|
||||
if (ALGO == Algorithm::CN_CCX) {
|
||||
if (props.isCCX()) {
|
||||
cx = _mm_load_si128(reinterpret_cast<const __m128i*>(&l0[interleaved_index<interleave>(idx0 & MASK)]));
|
||||
cryptonight_conceal_tweak(cx, conc_var);
|
||||
cx = soft_aesenc(&cx, ax0, reinterpret_cast<const uint32_t*>(saes_table));
|
||||
@@ -750,7 +740,7 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
cx = _mm_aesenc_si128(cx, ax0);
|
||||
}
|
||||
|
||||
if (BASE == Algorithm::CN_1 || BASE == Algorithm::CN_2) {
|
||||
if (props.isBase1() || props.isBase2()) {
|
||||
cryptonight_monero_tweak<ALGO>(reinterpret_cast<uint64_t*>(&l0[interleaved_index<interleave>(idx0 & MASK)]), l0, idx0 & MASK, ax0, bx0, bx1, cx);
|
||||
} else {
|
||||
_mm_store_si128(reinterpret_cast<__m128i *>(&l0[interleaved_index<interleave>(idx0 & MASK)]), _mm_xor_si128(bx0, cx));
|
||||
@@ -762,13 +752,11 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
cl = (reinterpret_cast<uint64_t*>(&l0[interleaved_index<interleave>(idx0 & MASK)]))[0];
|
||||
ch = (reinterpret_cast<uint64_t*>(&l0[interleaved_index<interleave>(idx0 & MASK)]))[1];
|
||||
|
||||
if (BASE == Algorithm::CN_2) {
|
||||
if (props.isBase2()) {
|
||||
if (props.isR()) {
|
||||
VARIANT4_RANDOM_MATH(0, al0, ah0, cl, bx0, bx1);
|
||||
if (ALGO == Algorithm::CN_R) {
|
||||
al0 ^= r0[2] | (static_cast<uint64_t>(r0[3]) << 32);
|
||||
ah0 ^= r0[0] | (static_cast<uint64_t>(r0[1]) << 32);
|
||||
}
|
||||
al0 ^= r0[2] | (static_cast<uint64_t>(r0[3]) << 32);
|
||||
ah0 ^= r0[0] | (static_cast<uint64_t>(r0[1]) << 32);
|
||||
} else {
|
||||
VARIANT2_INTEGER_MATH(0, cl, cx);
|
||||
}
|
||||
@@ -776,11 +764,11 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
lo = __umul128(idx0, cl, &hi);
|
||||
|
||||
if (BASE == Algorithm::CN_2) {
|
||||
if (ALGO == Algorithm::CN_R) {
|
||||
if (props.isBase2()) {
|
||||
if (props.isR()) {
|
||||
VARIANT2_SHUFFLE(l0, idx0 & MASK, ax0, bx0, bx1, cx, 0);
|
||||
} else {
|
||||
VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx0, bx1, hi, lo, (((ALGO == Algorithm::CN_RWZ) || (ALGO == Algorithm::CN_UPX2)) ? 1 : 0));
|
||||
VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx0, bx1, hi, lo, ((props.isRWZ() || props.isUPX2()) ? 1 : 0));
|
||||
}
|
||||
}
|
||||
|
||||
@@ -789,9 +777,9 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
reinterpret_cast<uint64_t*>(&l0[interleaved_index<interleave>(idx0 & MASK)])[0] = al0;
|
||||
|
||||
if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) {
|
||||
if (props.isHeavyTube() || props.isRTO()) {
|
||||
reinterpret_cast<uint64_t*>(&l0[interleaved_index<interleave>(idx0 & MASK)])[1] = ah0 ^ tweak1_2_0 ^ al0;
|
||||
} else if (BASE == Algorithm::CN_1) {
|
||||
} else if (props.isBase1()) {
|
||||
reinterpret_cast<uint64_t*>(&l0[interleaved_index<interleave>(idx0 & MASK)])[1] = ah0 ^ tweak1_2_0;
|
||||
} else {
|
||||
reinterpret_cast<uint64_t*>(&l0[interleaved_index<interleave>(idx0 & MASK)])[1] = ah0;
|
||||
@@ -819,7 +807,7 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
((int64_t*)&l0[interleaved_index<interleave>(idx0 & MASK)])[0] = n ^ q;
|
||||
|
||||
if (ALGO == Algorithm::CN_HEAVY_XHV) {
|
||||
if (props.isHeavyXHV()) {
|
||||
d = ~d;
|
||||
}
|
||||
|
||||
@@ -827,7 +815,7 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
|
||||
}
|
||||
# endif
|
||||
|
||||
if (BASE == Algorithm::CN_2) {
|
||||
if (props.isBase2()) {
|
||||
bx1 = bx0;
|
||||
}
|
||||
|
||||
@@ -960,7 +948,7 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
}
|
||||
cn_explode_scratchpad<ALGO, false, 0>(ctx[0]);
|
||||
|
||||
if (ALGO == Algorithm::CN_2) {
|
||||
if (props.is2()) {
|
||||
if (ASM == Assembly::INTEL) {
|
||||
cnv2_mainloop_ivybridge_asm(ctx);
|
||||
}
|
||||
@@ -971,7 +959,7 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
cnv2_mainloop_bulldozer_asm(ctx);
|
||||
}
|
||||
}
|
||||
else if (ALGO == Algorithm::CN_HALF) {
|
||||
else if (props.isHalf()) {
|
||||
if (ASM == Assembly::INTEL) {
|
||||
cn_half_mainloop_ivybridge_asm(ctx);
|
||||
}
|
||||
@@ -983,7 +971,7 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
}
|
||||
}
|
||||
# ifdef XMRIG_ALGO_CN_PICO
|
||||
else if (ALGO == Algorithm::CN_PICO_0) {
|
||||
else if (props.isPico0()) {
|
||||
if (ASM == Assembly::INTEL) {
|
||||
cn_trtl_mainloop_ivybridge_asm(ctx);
|
||||
}
|
||||
@@ -994,7 +982,7 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
cn_trtl_mainloop_bulldozer_asm(ctx);
|
||||
}
|
||||
}
|
||||
else if (ALGO == Algorithm::CN_PICO_TLO) {
|
||||
else if (props.isPicoTLO()) {
|
||||
if (ASM == Assembly::INTEL) {
|
||||
cn_tlo_mainloop_ivybridge_asm(ctx);
|
||||
}
|
||||
@@ -1006,10 +994,10 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
}
|
||||
}
|
||||
# endif
|
||||
else if (ALGO == Algorithm::CN_RWZ) {
|
||||
else if (props.isRWZ()) {
|
||||
cnv2_rwz_mainloop_asm(ctx);
|
||||
}
|
||||
else if (ALGO == Algorithm::CN_ZLS) {
|
||||
else if (props.isZLS()) {
|
||||
if (ASM == Assembly::INTEL) {
|
||||
cn_zls_mainloop_ivybridge_asm(ctx);
|
||||
}
|
||||
@@ -1020,7 +1008,7 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
cn_zls_mainloop_bulldozer_asm(ctx);
|
||||
}
|
||||
}
|
||||
else if (ALGO == Algorithm::CN_DOUBLE) {
|
||||
else if (props.isDouble()) {
|
||||
if (ASM == Assembly::INTEL) {
|
||||
cn_double_mainloop_ivybridge_asm(ctx);
|
||||
}
|
||||
@@ -1032,7 +1020,7 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
}
|
||||
}
|
||||
# ifdef XMRIG_ALGO_CN_FEMTO
|
||||
else if (ALGO == Algorithm::CN_UPX2) {
|
||||
else if (props.isUPX2()) {
|
||||
cn_upx2_mainloop_asm(ctx);
|
||||
}
|
||||
# endif
|
||||
@@ -1078,22 +1066,22 @@ inline void cryptonight_double_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
cn_explode_scratchpad<ALGO, false, 0>(ctx[1]);
|
||||
}
|
||||
|
||||
if (ALGO == Algorithm::CN_2) {
|
||||
if (props.is2()) {
|
||||
cnv2_double_mainloop_sandybridge_asm(ctx);
|
||||
}
|
||||
else if (ALGO == Algorithm::CN_HALF) {
|
||||
else if (props.isHalf()){
|
||||
cn_half_double_mainloop_sandybridge_asm(ctx);
|
||||
}
|
||||
# ifdef XMRIG_ALGO_CN_PICO
|
||||
else if (ALGO == Algorithm::CN_PICO_0) {
|
||||
else if (props.isPico0()) {
|
||||
cn_trtl_double_mainloop_sandybridge_asm(ctx);
|
||||
}
|
||||
else if (ALGO == Algorithm::CN_PICO_TLO) {
|
||||
else if (props.isPicoTLO()) {
|
||||
cn_tlo_double_mainloop_sandybridge_asm(ctx);
|
||||
}
|
||||
# endif
|
||||
# ifdef XMRIG_ALGO_CN_FEMTO
|
||||
else if (ALGO == Algorithm::CN_UPX2) {
|
||||
else if (props.isUPX2()) {
|
||||
if (Cpu::info()->arch() == ICpuInfo::ARCH_ZEN3) {
|
||||
cnv2_upx_double_mainloop_zen3_asm(ctx);
|
||||
}
|
||||
@@ -1102,13 +1090,13 @@ inline void cryptonight_double_hash_asm(const uint8_t *__restrict__ input, size_
|
||||
}
|
||||
}
|
||||
# endif
|
||||
else if (ALGO == Algorithm::CN_RWZ) {
|
||||
else if (props.isRWZ()) {
|
||||
cnv2_rwz_double_mainloop_asm(ctx);
|
||||
}
|
||||
else if (ALGO == Algorithm::CN_ZLS) {
|
||||
else if (props.isZLS()) {
|
||||
cn_zls_double_mainloop_sandybridge_asm(ctx);
|
||||
}
|
||||
else if (ALGO == Algorithm::CN_DOUBLE) {
|
||||
else if (props.isDouble()) {
|
||||
cn_double_double_mainloop_sandybridge_asm(ctx);
|
||||
}
|
||||
else if (props.isR()) {
|
||||
@@ -1146,9 +1134,8 @@ template<Algorithm::Id ALGO>
|
||||
static NOINLINE void cryptonight_single_hash_gr_sse41(const uint8_t* __restrict__ input, size_t size, uint8_t* __restrict__ output, cryptonight_ctx** __restrict__ ctx, uint64_t height)
|
||||
{
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
constexpr Algorithm::Id BASE = props.base();
|
||||
|
||||
if (BASE == Algorithm::CN_1 && size < 43) {
|
||||
if (props.isBase1() && size < 43) {
|
||||
memset(output, 0, 32);
|
||||
return;
|
||||
}
|
||||
@@ -1163,12 +1150,12 @@ static NOINLINE void cryptonight_single_hash_gr_sse41(const uint8_t* __restrict_
|
||||
VARIANT1_INIT(0);
|
||||
ctx[0]->tweak1_2 = tweak1_2_0;
|
||||
ctx[0]->tweak1_table = tweak1_table;
|
||||
if (ALGO == Algorithm::CN_GR_0) cn_gr0_single_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_1) cn_gr1_single_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_2) cn_gr2_single_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_3) cn_gr3_single_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_4) cn_gr4_single_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_5) cn_gr5_single_mainloop_asm(ctx);
|
||||
if (props.isGR0()) cn_gr0_single_mainloop_asm(ctx);
|
||||
if (props.isGR1()) cn_gr1_single_mainloop_asm(ctx);
|
||||
if (props.isGR2()) cn_gr2_single_mainloop_asm(ctx);
|
||||
if (props.isGR3()) cn_gr3_single_mainloop_asm(ctx);
|
||||
if (props.isGR4()) cn_gr4_single_mainloop_asm(ctx);
|
||||
if (props.isGR5()) cn_gr5_single_mainloop_asm(ctx);
|
||||
|
||||
cn_implode_scratchpad<ALGO, false, 0>(ctx[0]);
|
||||
keccakf(reinterpret_cast<uint64_t*>(ctx[0]->state), 24);
|
||||
@@ -1180,9 +1167,8 @@ template<Algorithm::Id ALGO>
|
||||
static NOINLINE void cryptonight_double_hash_gr_sse41(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height)
|
||||
{
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
constexpr Algorithm::Id BASE = props.base();
|
||||
|
||||
if (BASE == Algorithm::CN_1 && size < 43) {
|
||||
if (props.isBase1() && size < 43) {
|
||||
memset(output, 0, 64);
|
||||
return;
|
||||
}
|
||||
@@ -1196,7 +1182,7 @@ static NOINLINE void cryptonight_double_hash_gr_sse41(const uint8_t *__restrict_
|
||||
}
|
||||
|
||||
# ifdef XMRIG_VAES
|
||||
if (!props.isHeavy() && cn_vaes_enabled) {
|
||||
if (cn_vaes_enabled) {
|
||||
cn_explode_scratchpad_vaes_double(ctx[0], ctx[1], props.memory(), props.half_mem());
|
||||
}
|
||||
else
|
||||
@@ -1214,15 +1200,15 @@ static NOINLINE void cryptonight_double_hash_gr_sse41(const uint8_t *__restrict_
|
||||
|
||||
ctx[0]->tweak1_table = tweak1_table;
|
||||
|
||||
if (ALGO == Algorithm::CN_GR_0) cn_gr0_double_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_1) cn_gr1_double_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_2) cn_gr2_double_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_3) cn_gr3_double_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_4) cn_gr4_double_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_5) cn_gr5_double_mainloop_asm(ctx);
|
||||
if (props.isGR0()) cn_gr0_double_mainloop_asm(ctx);
|
||||
if (props.isGR1()) cn_gr1_double_mainloop_asm(ctx);
|
||||
if (props.isGR2()) cn_gr2_double_mainloop_asm(ctx);
|
||||
if (props.isGR3()) cn_gr3_double_mainloop_asm(ctx);
|
||||
if (props.isGR4()) cn_gr4_double_mainloop_asm(ctx);
|
||||
if (props.isGR5()) cn_gr5_double_mainloop_asm(ctx);
|
||||
|
||||
# ifdef XMRIG_VAES
|
||||
if (!props.isHeavy() && cn_vaes_enabled) {
|
||||
if (cn_vaes_enabled) {
|
||||
cn_implode_scratchpad_vaes_double(ctx[0], ctx[1], props.memory(), props.half_mem());
|
||||
}
|
||||
else
|
||||
@@ -1267,15 +1253,8 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
constexpr size_t MASK = props.mask();
|
||||
constexpr Algorithm::Id BASE = props.base();
|
||||
|
||||
# ifdef XMRIG_ALGO_CN_HEAVY
|
||||
constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE;
|
||||
# else
|
||||
constexpr bool IS_CN_HEAVY_TUBE = false;
|
||||
# endif
|
||||
|
||||
if (BASE == Algorithm::CN_1 && size < 43) {
|
||||
if (props.isBase1() && size < 43) {
|
||||
memset(output, 0, 64);
|
||||
return;
|
||||
}
|
||||
@@ -1323,7 +1302,7 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
__m128i bx11 = _mm_set_epi64x(h1[9] ^ h1[11], h1[8] ^ h1[10]);
|
||||
|
||||
__m128 conc_var0, conc_var1;
|
||||
if (ALGO == Algorithm::CN_CCX) {
|
||||
if (props.isCCX()) {
|
||||
conc_var0 = _mm_setzero_ps();
|
||||
conc_var1 = _mm_setzero_ps();
|
||||
RESTORE_ROUNDING_MODE();
|
||||
@@ -1334,10 +1313,10 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
for (size_t i = 0; i < props.iterations(); i++) {
|
||||
__m128i cx0, cx1;
|
||||
if (IS_CN_HEAVY_TUBE || !SOFT_AES) {
|
||||
if (props.isHeavyTube() || !SOFT_AES) {
|
||||
cx0 = _mm_load_si128(reinterpret_cast<const __m128i *>(&l0[idx0 & MASK]));
|
||||
cx1 = _mm_load_si128(reinterpret_cast<const __m128i *>(&l1[idx1 & MASK]));
|
||||
if (ALGO == Algorithm::CN_CCX) {
|
||||
if (props.isCCX()) {
|
||||
cryptonight_conceal_tweak(cx0, conc_var0);
|
||||
cryptonight_conceal_tweak(cx1, conc_var1);
|
||||
}
|
||||
@@ -1345,12 +1324,12 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
const __m128i ax0 = _mm_set_epi64x(ah0, al0);
|
||||
const __m128i ax1 = _mm_set_epi64x(ah1, al1);
|
||||
if (IS_CN_HEAVY_TUBE) {
|
||||
if (props.isHeavyTube()) {
|
||||
cx0 = aes_round_tweak_div(cx0, ax0);
|
||||
cx1 = aes_round_tweak_div(cx1, ax1);
|
||||
}
|
||||
else if (SOFT_AES) {
|
||||
if (ALGO == Algorithm::CN_CCX) {
|
||||
if (props.isCCX()) {
|
||||
cx0 = _mm_load_si128(reinterpret_cast<const __m128i*>(&l0[idx0 & MASK]));
|
||||
cx1 = _mm_load_si128(reinterpret_cast<const __m128i*>(&l1[idx1 & MASK]));
|
||||
cryptonight_conceal_tweak(cx0, conc_var0);
|
||||
@@ -1368,7 +1347,7 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
cx1 = _mm_aesenc_si128(cx1, ax1);
|
||||
}
|
||||
|
||||
if (BASE == Algorithm::CN_1 || BASE == Algorithm::CN_2) {
|
||||
if (props.isBase1() || props.isBase2()) {
|
||||
cryptonight_monero_tweak<ALGO>((uint64_t*)&l0[idx0 & MASK], l0, idx0 & MASK, ax0, bx00, bx01, cx0);
|
||||
cryptonight_monero_tweak<ALGO>((uint64_t*)&l1[idx1 & MASK], l1, idx1 & MASK, ax1, bx10, bx11, cx1);
|
||||
} else {
|
||||
@@ -1383,13 +1362,11 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
cl = ((uint64_t*) &l0[idx0 & MASK])[0];
|
||||
ch = ((uint64_t*) &l0[idx0 & MASK])[1];
|
||||
|
||||
if (BASE == Algorithm::CN_2) {
|
||||
if (props.isBase2()) {
|
||||
if (props.isR()) {
|
||||
VARIANT4_RANDOM_MATH(0, al0, ah0, cl, bx00, bx01);
|
||||
if (ALGO == Algorithm::CN_R) {
|
||||
al0 ^= r0[2] | ((uint64_t)(r0[3]) << 32);
|
||||
ah0 ^= r0[0] | ((uint64_t)(r0[1]) << 32);
|
||||
}
|
||||
al0 ^= r0[2] | ((uint64_t)(r0[3]) << 32);
|
||||
ah0 ^= r0[0] | ((uint64_t)(r0[1]) << 32);
|
||||
} else {
|
||||
VARIANT2_INTEGER_MATH(0, cl, cx0);
|
||||
}
|
||||
@@ -1397,11 +1374,11 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
lo = __umul128(idx0, cl, &hi);
|
||||
|
||||
if (BASE == Algorithm::CN_2) {
|
||||
if (ALGO == Algorithm::CN_R) {
|
||||
if (props.isBase2()) {
|
||||
if (props.isR()) {
|
||||
VARIANT2_SHUFFLE(l0, idx0 & MASK, ax0, bx00, bx01, cx0, 0);
|
||||
} else {
|
||||
VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx00, bx01, hi, lo, (((ALGO == Algorithm::CN_RWZ) || (ALGO == Algorithm::CN_UPX2)) ? 1 : 0));
|
||||
VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx00, bx01, hi, lo, ((props.isRWZ() || props.isUPX2()) ? 1 : 0));
|
||||
}
|
||||
}
|
||||
|
||||
@@ -1410,9 +1387,9 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
((uint64_t*)&l0[idx0 & MASK])[0] = al0;
|
||||
|
||||
if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) {
|
||||
if (props.isHeavyTube() || props.isRTO()) {
|
||||
((uint64_t*) &l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0 ^ al0;
|
||||
} else if (BASE == Algorithm::CN_1) {
|
||||
} else if (props.isBase1()) {
|
||||
((uint64_t*) &l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0;
|
||||
} else {
|
||||
((uint64_t*) &l0[idx0 & MASK])[1] = ah0;
|
||||
@@ -1430,7 +1407,7 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
((int64_t*)&l0[idx0 & MASK])[0] = n ^ q;
|
||||
|
||||
if (ALGO == Algorithm::CN_HEAVY_XHV) {
|
||||
if (props.isHeavyXHV()) {
|
||||
d = ~d;
|
||||
}
|
||||
|
||||
@@ -1441,13 +1418,11 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
cl = ((uint64_t*) &l1[idx1 & MASK])[0];
|
||||
ch = ((uint64_t*) &l1[idx1 & MASK])[1];
|
||||
|
||||
if (BASE == Algorithm::CN_2) {
|
||||
if (props.isBase2()) {
|
||||
if (props.isR()) {
|
||||
VARIANT4_RANDOM_MATH(1, al1, ah1, cl, bx10, bx11);
|
||||
if (ALGO == Algorithm::CN_R) {
|
||||
al1 ^= r1[2] | ((uint64_t)(r1[3]) << 32);
|
||||
ah1 ^= r1[0] | ((uint64_t)(r1[1]) << 32);
|
||||
}
|
||||
al1 ^= r1[2] | ((uint64_t)(r1[3]) << 32);
|
||||
ah1 ^= r1[0] | ((uint64_t)(r1[1]) << 32);
|
||||
} else {
|
||||
VARIANT2_INTEGER_MATH(1, cl, cx1);
|
||||
}
|
||||
@@ -1455,11 +1430,11 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
lo = __umul128(idx1, cl, &hi);
|
||||
|
||||
if (BASE == Algorithm::CN_2) {
|
||||
if (ALGO == Algorithm::CN_R) {
|
||||
if (props.isBase2()) {
|
||||
if (props.isR()) {
|
||||
VARIANT2_SHUFFLE(l1, idx1 & MASK, ax1, bx10, bx11, cx1, 0);
|
||||
} else {
|
||||
VARIANT2_SHUFFLE2(l1, idx1 & MASK, ax1, bx10, bx11, hi, lo, (((ALGO == Algorithm::CN_RWZ) || (ALGO == Algorithm::CN_UPX2)) ? 1 : 0));
|
||||
VARIANT2_SHUFFLE2(l1, idx1 & MASK, ax1, bx10, bx11, hi, lo, ((props.isRWZ() || props.isUPX2()) ? 1 : 0));
|
||||
}
|
||||
}
|
||||
|
||||
@@ -1468,9 +1443,9 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
((uint64_t*)&l1[idx1 & MASK])[0] = al1;
|
||||
|
||||
if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) {
|
||||
if (props.isHeavyTube() || props.isRTO()) {
|
||||
((uint64_t*)&l1[idx1 & MASK])[1] = ah1 ^ tweak1_2_1 ^ al1;
|
||||
} else if (BASE == Algorithm::CN_1) {
|
||||
} else if (props.isBase1()) {
|
||||
((uint64_t*)&l1[idx1 & MASK])[1] = ah1 ^ tweak1_2_1;
|
||||
} else {
|
||||
((uint64_t*)&l1[idx1 & MASK])[1] = ah1;
|
||||
@@ -1488,7 +1463,7 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
|
||||
((int64_t*)&l1[idx1 & MASK])[0] = n ^ q;
|
||||
|
||||
if (ALGO == Algorithm::CN_HEAVY_XHV) {
|
||||
if (props.isHeavyXHV()) {
|
||||
d = ~d;
|
||||
}
|
||||
|
||||
@@ -1496,7 +1471,7 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si
|
||||
}
|
||||
# endif
|
||||
|
||||
if (BASE == Algorithm::CN_2) {
|
||||
if (props.isBase2()) {
|
||||
bx01 = bx00;
|
||||
bx11 = bx10;
|
||||
}
|
||||
@@ -1529,9 +1504,8 @@ template<Algorithm::Id ALGO>
|
||||
static NOINLINE void cryptonight_quad_hash_gr_sse41(const uint8_t* __restrict__ input, size_t size, uint8_t* __restrict__ output, cryptonight_ctx** __restrict__ ctx, uint64_t height)
|
||||
{
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
constexpr Algorithm::Id BASE = props.base();
|
||||
|
||||
if (BASE == Algorithm::CN_1 && size < 43) {
|
||||
if (props.isBase1() && size < 43) {
|
||||
memset(output, 0, 32 * 4);
|
||||
return;
|
||||
}
|
||||
@@ -1549,7 +1523,7 @@ static NOINLINE void cryptonight_quad_hash_gr_sse41(const uint8_t* __restrict__
|
||||
}
|
||||
|
||||
# ifdef XMRIG_VAES
|
||||
if (!props.isHeavy() && cn_vaes_enabled) {
|
||||
if (cn_vaes_enabled) {
|
||||
cn_explode_scratchpad_vaes_double(ctx[0], ctx[1], props.memory(), props.half_mem());
|
||||
cn_explode_scratchpad_vaes_double(ctx[2], ctx[3], props.memory(), props.half_mem());
|
||||
}
|
||||
@@ -1569,15 +1543,15 @@ static NOINLINE void cryptonight_quad_hash_gr_sse41(const uint8_t* __restrict__
|
||||
|
||||
ctx[0]->tweak1_table = tweak1_table;
|
||||
|
||||
if (ALGO == Algorithm::CN_GR_0) cn_gr0_quad_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_1) cn_gr1_quad_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_2) cn_gr2_quad_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_3) cn_gr3_quad_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_4) cn_gr4_quad_mainloop_asm(ctx);
|
||||
if (ALGO == Algorithm::CN_GR_5) cn_gr5_quad_mainloop_asm(ctx);
|
||||
if (props.isGR0()) cn_gr0_quad_mainloop_asm(ctx);
|
||||
if (props.isGR1()) cn_gr1_quad_mainloop_asm(ctx);
|
||||
if (props.isGR2()) cn_gr2_quad_mainloop_asm(ctx);
|
||||
if (props.isGR3()) cn_gr3_quad_mainloop_asm(ctx);
|
||||
if (props.isGR4()) cn_gr4_quad_mainloop_asm(ctx);
|
||||
if (props.isGR5()) cn_gr5_quad_mainloop_asm(ctx);
|
||||
|
||||
# ifdef XMRIG_VAES
|
||||
if (!props.isHeavy() && cn_vaes_enabled) {
|
||||
if (cn_vaes_enabled) {
|
||||
cn_implode_scratchpad_vaes_double(ctx[0], ctx[1], props.memory(), props.half_mem());
|
||||
cn_implode_scratchpad_vaes_double(ctx[2], ctx[3], props.memory(), props.half_mem());
|
||||
}
|
||||
@@ -1606,14 +1580,14 @@ static NOINLINE void cryptonight_quad_hash_gr_sse41(const uint8_t* __restrict__
|
||||
#define CN_STEP1(a, b0, b1, c, l, ptr, idx, conc_var) \
|
||||
ptr = reinterpret_cast<__m128i*>(&l[idx & MASK]); \
|
||||
c = _mm_load_si128(ptr); \
|
||||
if (ALGO == Algorithm::CN_CCX) { \
|
||||
if (props.isCCX()) { \
|
||||
cryptonight_conceal_tweak(c, conc_var); \
|
||||
}
|
||||
|
||||
|
||||
|
||||
#define CN_STEP2(a, b0, b1, c, l, ptr, idx) \
|
||||
if (IS_CN_HEAVY_TUBE) { \
|
||||
if (props.isHeavyTube()) { \
|
||||
c = aes_round_tweak_div(c, a); \
|
||||
} \
|
||||
else if (SOFT_AES) { \
|
||||
@@ -1622,7 +1596,7 @@ static NOINLINE void cryptonight_quad_hash_gr_sse41(const uint8_t* __restrict__
|
||||
c = _mm_aesenc_si128(c, a); \
|
||||
} \
|
||||
\
|
||||
if (BASE == Algorithm::CN_1 || BASE == Algorithm::CN_2) { \
|
||||
if (props.isBase1() || props.isBase2()) { \
|
||||
cryptonight_monero_tweak<ALGO>((uint64_t*)ptr, l, idx & MASK, a, b0, b1, c); \
|
||||
} else { \
|
||||
_mm_store_si128(ptr, _mm_xor_si128(b0, c)); \
|
||||
@@ -1638,36 +1612,34 @@ static NOINLINE void cryptonight_quad_hash_gr_sse41(const uint8_t* __restrict__
|
||||
|
||||
#define CN_STEP4(part, a, b0, b1, c, l, mc, ptr, idx) \
|
||||
uint64_t al##part, ah##part; \
|
||||
if (BASE == Algorithm::CN_2) { \
|
||||
if (props.isBase2()) { \
|
||||
if (props.isR()) { \
|
||||
al##part = _mm_cvtsi128_si64(a); \
|
||||
ah##part = _mm_cvtsi128_si64(_mm_srli_si128(a, 8)); \
|
||||
VARIANT4_RANDOM_MATH(part, al##part, ah##part, cl##part, b0, b1); \
|
||||
if (ALGO == Algorithm::CN_R) { \
|
||||
al##part ^= r##part[2] | ((uint64_t)(r##part[3]) << 32); \
|
||||
ah##part ^= r##part[0] | ((uint64_t)(r##part[1]) << 32); \
|
||||
} \
|
||||
al##part ^= r##part[2] | ((uint64_t)(r##part[3]) << 32); \
|
||||
ah##part ^= r##part[0] | ((uint64_t)(r##part[1]) << 32); \
|
||||
} else { \
|
||||
VARIANT2_INTEGER_MATH(part, cl##part, c); \
|
||||
} \
|
||||
} \
|
||||
lo = __umul128(idx, cl##part, &hi); \
|
||||
if (BASE == Algorithm::CN_2) { \
|
||||
if (ALGO == Algorithm::CN_R) { \
|
||||
if (props.isBase2()) { \
|
||||
if (props.isR()) { \
|
||||
VARIANT2_SHUFFLE(l, idx & MASK, a, b0, b1, c, 0); \
|
||||
} else { \
|
||||
VARIANT2_SHUFFLE2(l, idx & MASK, a, b0, b1, hi, lo, (((ALGO == Algorithm::CN_RWZ) || (ALGO == Algorithm::CN_UPX2)) ? 1 : 0)); \
|
||||
VARIANT2_SHUFFLE2(l, idx & MASK, a, b0, b1, hi, lo, ((props.isRWZ() || props.isUPX2()) ? 1 : 0)); \
|
||||
} \
|
||||
} \
|
||||
if (ALGO == Algorithm::CN_R) { \
|
||||
if (props.isR()) { \
|
||||
a = _mm_set_epi64x(ah##part, al##part); \
|
||||
} \
|
||||
a = _mm_add_epi64(a, _mm_set_epi64x(lo, hi)); \
|
||||
\
|
||||
if (BASE == Algorithm::CN_1) { \
|
||||
if (props.isBase1()) { \
|
||||
_mm_store_si128(ptr, _mm_xor_si128(a, mc)); \
|
||||
\
|
||||
if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) { \
|
||||
if (props.isHeavyTube() || props.isRTO()) { \
|
||||
((uint64_t*)ptr)[1] ^= ((uint64_t*)ptr)[0]; \
|
||||
} \
|
||||
} else { \
|
||||
@@ -1681,13 +1653,13 @@ static NOINLINE void cryptonight_quad_hash_gr_sse41(const uint8_t* __restrict__
|
||||
int32_t d = ((int32_t*)&l[idx & MASK])[2]; \
|
||||
int64_t q = n / (d | 0x5); \
|
||||
((int64_t*)&l[idx & MASK])[0] = n ^ q; \
|
||||
if (IS_CN_HEAVY_XHV) { \
|
||||
if (props.isHeavyXHV()) { \
|
||||
d = ~d; \
|
||||
} \
|
||||
\
|
||||
idx = d ^ q; \
|
||||
} \
|
||||
if (BASE == Algorithm::CN_2) { \
|
||||
if (props.isBase2()) { \
|
||||
b1 = b0; \
|
||||
} \
|
||||
b0 = c;
|
||||
@@ -1697,11 +1669,11 @@ static NOINLINE void cryptonight_quad_hash_gr_sse41(const uint8_t* __restrict__
|
||||
__m128i mc##n; \
|
||||
__m128i division_result_xmm_##n; \
|
||||
__m128i sqrt_result_xmm_##n; \
|
||||
if (BASE == Algorithm::CN_1) { \
|
||||
if (props.isBase1()) { \
|
||||
mc##n = _mm_set_epi64x(*reinterpret_cast<const uint64_t*>(input + n * size + 35) ^ \
|
||||
*(reinterpret_cast<const uint64_t*>((ctx)->state) + 24), 0); \
|
||||
} \
|
||||
if (BASE == Algorithm::CN_2) { \
|
||||
if (props.isBase2()) { \
|
||||
division_result_xmm_##n = _mm_cvtsi64_si128(h##n[12]); \
|
||||
sqrt_result_xmm_##n = _mm_cvtsi64_si128(h##n[13]); \
|
||||
} \
|
||||
@@ -1710,7 +1682,7 @@ static NOINLINE void cryptonight_quad_hash_gr_sse41(const uint8_t* __restrict__
|
||||
__m128i bx##n##1 = _mm_set_epi64x(h##n[9] ^ h##n[11], h##n[8] ^ h##n[10]); \
|
||||
__m128i cx##n = _mm_setzero_si128(); \
|
||||
__m128 conc_var##n; \
|
||||
if (ALGO == Algorithm::CN_CCX) { \
|
||||
if (props.isCCX()) { \
|
||||
conc_var##n = _mm_setzero_ps(); \
|
||||
} \
|
||||
VARIANT4_RANDOM_MATH_INIT(n);
|
||||
@@ -1721,17 +1693,8 @@ inline void cryptonight_triple_hash(const uint8_t *__restrict__ input, size_t si
|
||||
{
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
constexpr size_t MASK = props.mask();
|
||||
constexpr Algorithm::Id BASE = props.base();
|
||||
|
||||
# ifdef XMRIG_ALGO_CN_HEAVY
|
||||
constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE;
|
||||
constexpr bool IS_CN_HEAVY_XHV = ALGO == Algorithm::CN_HEAVY_XHV;
|
||||
# else
|
||||
constexpr bool IS_CN_HEAVY_TUBE = false;
|
||||
constexpr bool IS_CN_HEAVY_XHV = false;
|
||||
# endif
|
||||
|
||||
if (BASE == Algorithm::CN_1 && size < 43) {
|
||||
if (props.isBase1() && size < 43) {
|
||||
memset(output, 0, 32 * 3);
|
||||
return;
|
||||
}
|
||||
@@ -1755,7 +1718,7 @@ inline void cryptonight_triple_hash(const uint8_t *__restrict__ input, size_t si
|
||||
CONST_INIT(ctx[1], 1);
|
||||
CONST_INIT(ctx[2], 2);
|
||||
VARIANT2_SET_ROUNDING_MODE();
|
||||
if (ALGO == Algorithm::CN_CCX) {
|
||||
if (props.isCCX()) {
|
||||
RESTORE_ROUNDING_MODE();
|
||||
}
|
||||
|
||||
@@ -1819,17 +1782,8 @@ inline void cryptonight_quad_hash(const uint8_t *__restrict__ input, size_t size
|
||||
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
constexpr size_t MASK = props.mask();
|
||||
constexpr Algorithm::Id BASE = props.base();
|
||||
|
||||
# ifdef XMRIG_ALGO_CN_HEAVY
|
||||
constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE;
|
||||
constexpr bool IS_CN_HEAVY_XHV = ALGO == Algorithm::CN_HEAVY_XHV;
|
||||
# else
|
||||
constexpr bool IS_CN_HEAVY_TUBE = false;
|
||||
constexpr bool IS_CN_HEAVY_XHV = false;
|
||||
# endif
|
||||
|
||||
if (BASE == Algorithm::CN_1 && size < 43) {
|
||||
if (props.isBase1() && size < 43) {
|
||||
memset(output, 0, 32 * 4);
|
||||
return;
|
||||
}
|
||||
@@ -1869,7 +1823,7 @@ inline void cryptonight_quad_hash(const uint8_t *__restrict__ input, size_t size
|
||||
CONST_INIT(ctx[2], 2);
|
||||
CONST_INIT(ctx[3], 3);
|
||||
VARIANT2_SET_ROUNDING_MODE();
|
||||
if (ALGO == Algorithm::CN_CCX) {
|
||||
if (props.isCCX()) {
|
||||
RESTORE_ROUNDING_MODE();
|
||||
}
|
||||
|
||||
@@ -1930,17 +1884,8 @@ inline void cryptonight_penta_hash(const uint8_t *__restrict__ input, size_t siz
|
||||
{
|
||||
constexpr CnAlgo<ALGO> props;
|
||||
constexpr size_t MASK = props.mask();
|
||||
constexpr Algorithm::Id BASE = props.base();
|
||||
|
||||
# ifdef XMRIG_ALGO_CN_HEAVY
|
||||
constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE;
|
||||
constexpr bool IS_CN_HEAVY_XHV = ALGO == Algorithm::CN_HEAVY_XHV;
|
||||
# else
|
||||
constexpr bool IS_CN_HEAVY_TUBE = false;
|
||||
constexpr bool IS_CN_HEAVY_XHV = false;
|
||||
# endif
|
||||
|
||||
if (BASE == Algorithm::CN_1 && size < 43) {
|
||||
if (props.isBase1() && size < 43) {
|
||||
memset(output, 0, 32 * 5);
|
||||
return;
|
||||
}
|
||||
@@ -1970,7 +1915,7 @@ inline void cryptonight_penta_hash(const uint8_t *__restrict__ input, size_t siz
|
||||
CONST_INIT(ctx[3], 3);
|
||||
CONST_INIT(ctx[4], 4);
|
||||
VARIANT2_SET_ROUNDING_MODE();
|
||||
if (ALGO == Algorithm::CN_CCX) {
|
||||
if (props.isCCX()) {
|
||||
RESTORE_ROUNDING_MODE();
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user