void argon2hash(void *output, const void *input)
{
unsigned int t_costs = 2;
unsigned int m_costs = 16;
unsigned int Nfactor = m_costs/2;
const uint32_t r = SCRYPT_R;
const uint32_t p = SCRYPT_P;
scrypt_aligned_alloc YX, V;
uint8_t *X, *Y;
uint32_t N, chunk_bytes, i;
N = (1 << (Nfactor + 1));
chunk_bytes = SCRYPT_BLOCK_BYTES * r * 2;
V = scrypt_alloc((uint64_t)N * chunk_bytes);
YX = scrypt_alloc((p + 1) * chunk_bytes);
/* 1: X = PBKDF2(password, salt) */
Y = YX.ptr;
X = Y + chunk_bytes;
argon2_hash(output, input, t_costs, m_costs, N, X, Y, V.ptr, p, r);
scrypt_free(&V);
scrypt_free(&YX);
}
/* simple cpu test (util.c) */
void scryptjanehash(void *output, const void *input )
{
scrypt_aligned_alloc YX, V;
uint8_t *X, *Y;
uint32_t chunk_bytes;
uint32_t N = (1 << ( opt_scrypt_n + 1));
const uint32_t r = SCRYPT_R;
const uint32_t p = SCRYPT_P;
memset(output, 0, 32);
chunk_bytes = SCRYPT_BLOCK_BYTES * r * 2;
if (!scrypt_alloc((uint64_t)N * chunk_bytes, &V)) return;
if (!scrypt_alloc((p + 1) * chunk_bytes, &YX)) {
scrypt_free(&V);
return;
}
Y = YX.ptr;
X = Y + chunk_bytes;
scrypt_N_1_1((unsigned char*)input, 80, (unsigned char*)input, 80,
N, (unsigned char*)output, 32, X, Y, V.ptr);
scrypt_free(&V);
scrypt_free(&YX);
}
void
scrypt(const uint8_t *password, size_t password_len, const uint8_t *salt, size_t salt_len, uint8_t Nfactor, uint8_t rfactor, uint8_t pfactor, uint8_t *out, size_t bytes) {
scrypt_aligned_alloc YX, V;
uint8_t *X, *Y;
uint32_t N, r, p, chunk_bytes, i;
#if !defined(SCRYPT_CHOOSE_COMPILETIME)
scrypt_ROMixfn scrypt_ROMix = scrypt_getROMix();
#endif
#if !defined(SCRYPT_TEST)
static int power_on_self_test = 0;
if (!power_on_self_test) {
power_on_self_test = 1;
if (!scrypt_power_on_self_test())
scrypt_fatal_error("scrypt: power on self test failed");
}
#endif
if (Nfactor > scrypt_maxN)
scrypt_fatal_error("scrypt: N out of range");
if (rfactor > scrypt_maxr)
scrypt_fatal_error("scrypt: r out of range");
if (pfactor > scrypt_maxp)
scrypt_fatal_error("scrypt: p out of range");
N = (1 << (Nfactor + 1));
r = (1 << rfactor);
p = (1 << pfactor);
chunk_bytes = SCRYPT_BLOCK_BYTES * r * 2;
V = scrypt_alloc((uint64_t)N * chunk_bytes);
YX = scrypt_alloc((p + 1) * chunk_bytes);
//printf("%d-%d-%d\n" ,N,r,p);
//printf("%d\n", chunk_bytes);
//printf("%d\n", (uint64_t)N * chunk_bytes);
/* 1: X = PBKDF2(password, salt) */
Y = YX.ptr;
X = Y + chunk_bytes;
scrypt_pbkdf2(password, password_len, salt, salt_len, 1, X, chunk_bytes * p);
/* 2: X = ROMix(X) */
for (i = 0; i < p; i++)
scrypt_ROMix((scrypt_mix_word_t *)(X + (chunk_bytes * i)), (scrypt_mix_word_t *)Y, (scrypt_mix_word_t *)V.ptr, N, r);
/* 3: Out = PBKDF2(password, X) */
scrypt_pbkdf2(password, password_len, X, chunk_bytes * p, 1, out, bytes);
scrypt_ensure_zero(YX.ptr, (p + 1) * chunk_bytes);
scrypt_free(&V);
scrypt_free(&YX);
}
static int
scrypt_power_on_self_test() {
const scrypt_test_setting *t;
uint8_t test_digest[64];
uint32_t i;
int res = 7, scrypt_valid;
scrypt_aligned_alloc YX, V;
uint8_t *X, *Y;
uint32_t N, chunk_bytes;
const uint32_t r = SCRYPT_R;
const uint32_t p = SCRYPT_P;
if (!scrypt_test_mix()) {
#if !defined(SCRYPT_TEST)
scrypt_fatal_error("scrypt: mix function power-on-self-test failed");
#endif
res &= ~1;
}
if (!scrypt_test_hash()) {
#if !defined(SCRYPT_TEST)
scrypt_fatal_error("scrypt: hash function power-on-self-test failed");
#endif
res &= ~2;
}
for (i = 0, scrypt_valid = 1; post_settings[i].pw; i++) {
t = post_settings + i;
N = (1 << (t->Nfactor + 1));
chunk_bytes = SCRYPT_BLOCK_BYTES * r * 2;
V = scrypt_alloc((uint64_t)N * chunk_bytes);
YX = scrypt_alloc((p + 1) * chunk_bytes);
Y = YX.ptr;
X = Y + chunk_bytes;
scrypt_N_1_1((uint8_t *)t->pw, strlen(t->pw), (uint8_t *)t->salt, strlen(t->salt), N, test_digest, sizeof(test_digest), X, Y, V.ptr);
scrypt_valid &= scrypt_verify(post_vectors[i], test_digest, sizeof(test_digest));
scrypt_free(&V);
scrypt_free(&YX);
}
if (!scrypt_valid) {
#if !defined(SCRYPT_TEST)
scrypt_fatal_error("scrypt: scrypt power-on-self-test failed");
#endif
res &= ~4;
}
return res;
}
void
my_scrypt(const uint8_t *password, size_t password_len, const uint8_t *salt, size_t salt_len, uint8_t *out) {
scrypt_aligned_alloc YX, V;
uint8_t *X, *Y;
#if !defined(SCRYPT_CHOOSE_COMPILETIME)
scrypt_ROMixfn scrypt_ROMix = scrypt_getROMix();
#endif
/*
#if !defined(SCRYPT_TEST)
static int power_on_self_test = 0;
if (!power_on_self_test) {
power_on_self_test = 1;
if (!scrypt_power_on_self_test())
scrypt_fatal_error("scrypt: power on self test failed");
}
#endif
*/
V = scrypt_alloc((uint64_t)512 * chunk_bytes);
YX = scrypt_alloc(2 * chunk_bytes);
/* 1: X = PBKDF2(password, salt) */
Y = YX.ptr;
X = Y + chunk_bytes;
scrypt_pbkdf2(password, password_len, salt, salt_len, 1, X, chunk_bytes);
/* 2: X = ROMix(X) */
scrypt_ROMix((scrypt_mix_word_t *)X, (scrypt_mix_word_t *)Y, (scrypt_mix_word_t *)V.ptr, 512, 1);
/* 3: Out = PBKDF2(password, X) */
scrypt_pbkdf2(password, password_len, X, chunk_bytes, 1, out, 32);
scrypt_ensure_zero(YX.ptr, 2 * chunk_bytes);
scrypt_free(&V);
scrypt_free(&YX);
}
int scanhash_argon2(int thr_id, uint32_t *pdata, const uint32_t *ptarget,
uint32_t max_nonce, uint64_t *hashes_done)
{
uint32_t n = pdata[19] - 1;
const uint32_t first_nonce = pdata[19];
const uint32_t Htarg = ptarget[7];
uint32_t hash64[8] __attribute__((aligned(32)));
uint32_t endiandata[32];
scrypt_aligned_alloc YX, V;
uint8_t *X, *Y;
uint32_t N, chunk_bytes;
unsigned int t_costs = 2;
unsigned int m_costs = 16;
unsigned int Nfactor = m_costs/2;
const uint32_t r = SCRYPT_R;
const uint32_t p = SCRYPT_P;
int i;
uint64_t htmax[] = {
0,
0xF,
0xFF,
0xFFF,
0xFFFF,
0xFFFFF,
0xFFFFFF,
0xFFFFFFF,
0x10000000
};
uint32_t masks[] = {
0xFFFFFFFF,
0xFFFFFFF0,
0xFFFFFF00,
0xFFFFF000,
0xFFFF0000,
0xFFF00000,
0xFF000000,
0xF0000000,
0
};
// we need bigendian data...
for (int kk=0; kk < 32; kk++) {
be32enc(&endiandata[kk], (pdata)[kk]);
};
N = (1 << (Nfactor + 1));
chunk_bytes = SCRYPT_BLOCK_BYTES * r * 2;
V = scrypt_alloc((uint64_t)N * chunk_bytes);
YX = scrypt_alloc((p + 1) * chunk_bytes);
Y = YX.ptr;
X = Y + chunk_bytes;
#ifdef DEBUG_ALGO
printf("[%d] Htarg=%X\n", thr_id, Htarg);
#endif
for (int m=0; m < sizeof(masks); m++) {
if (Htarg <= htmax[m]) {
uint32_t mask = masks[m];
do {
pdata[19] = ++n;
be32enc(&endiandata[19], n);
argon2_hash(hash64, endiandata, t_costs, m_costs, N, X, Y, V.ptr, p, r);
#ifndef DEBUG_ALGO
if ((!(hash64[7] & mask)) && fulltest(hash64, ptarget)) {
*hashes_done = n - first_nonce + 1;
pdata[19] = n;
scrypt_free(&V);
scrypt_free(&YX);
return 1;
}
#else
if (!(n % 0x1000) && !thr_id) printf(".");
if (!(hash64[7] & mask)) {
printf("[%d]",thr_id);
if (fulltest(hash64, ptarget)) {
*hashes_done = n - first_nonce + 1;
scrypt_free(&V);
scrypt_free(&YX);
return true;
}
}
#endif
} while (n < max_nonce && !work_restart[thr_id].restart);
break;
}
}
scrypt_free(&V);
scrypt_free(&YX);
*hashes_done = n - first_nonce + 1;
pdata[19] = n;
return 0;
}
int scanhash_scrypt_jane(int thr_id, uint32_t *pdata,
const uint32_t *ptarget,
uint32_t max_nonce, struct timeval *tv_start, struct timeval *tv_end, unsigned long *hashes_done)
{
const uint32_t Htarg = ptarget[7];
static int s_Nfactor = 0;
if (s_Nfactor == 0 && strlen(jane_params) > 0)
applog(LOG_INFO, "Given scrypt-jane parameters: %s", jane_params);
int Nfactor = GetNfactor(bswap_32x4(pdata[17]));
if (Nfactor > scrypt_maxN) {
scrypt_fatal_error("scrypt: N out of range");
}
if (Nfactor != s_Nfactor)
{
// all of this isn't very thread-safe...
N = (1 << (Nfactor + 1));
applog(LOG_INFO, "Nfactor is %d (N=%d)!", Nfactor, N);
if (s_Nfactor != 0) {
// handle N-factor increase at runtime
// by adjusting the lookup_gap by factor 2
if (s_Nfactor == Nfactor-1)
for (int i=0; i < 8; ++i)
device_lookup_gap[i] *= 2;
}
s_Nfactor = Nfactor;
}
int throughput = cuda_throughput(thr_id);
if(throughput == 0)
return -1;
gettimeofday(tv_start, NULL);
uint32_t *data[2] = { new uint32_t[20*throughput], new uint32_t[20*throughput] };
uint32_t* hash[2] = { cuda_hashbuffer(thr_id,0), cuda_hashbuffer(thr_id,1) };
uint32_t n = pdata[19];
/* byte swap pdata into data[0]/[1] arrays */
for (int k=0; k<2; ++k) {
for(int z=0;z<20;z++) data[k][z] = bswap_32x4(pdata[z]);
for(int i=1;i<throughput;++i) memcpy(&data[k][20*i], &data[k][0], 20*sizeof(uint32_t));
}
if (parallel == 2) prepare_keccak512(thr_id, pdata);
scrypt_aligned_alloc Xbuf[2] = { scrypt_alloc(128 * throughput), scrypt_alloc(128 * throughput) };
scrypt_aligned_alloc Vbuf = scrypt_alloc((uint64_t)N * 128);
scrypt_aligned_alloc Ybuf = scrypt_alloc(128);
uint32_t nonce[2];
uint32_t* cuda_X[2] = { cuda_transferbuffer(thr_id,0), cuda_transferbuffer(thr_id,1) };
#if !defined(SCRYPT_CHOOSE_COMPILETIME)
scrypt_ROMixfn scrypt_ROMix = scrypt_getROMix();
#endif
int cur = 0, nxt = 1;
int iteration = 0;
do {
nonce[nxt] = n;
if (parallel < 2)
{
for(int i=0;i<throughput;++i) {
uint32_t tmp_nonce = n++;
data[nxt][20*i + 19] = bswap_32x4(tmp_nonce);
}
for(int i=0;i<throughput;++i)
scrypt_pbkdf2_1((unsigned char *)&data[nxt][20*i], 80, (unsigned char *)&data[nxt][20*i], 80, Xbuf[nxt].ptr + 128 * i, 128);
memcpy(cuda_X[nxt], Xbuf[nxt].ptr, 128 * throughput);
cuda_scrypt_serialize(thr_id, nxt);
cuda_scrypt_HtoD(thr_id, cuda_X[nxt], nxt);
cuda_scrypt_core(thr_id, nxt, N);
cuda_scrypt_done(thr_id, nxt);
cuda_scrypt_DtoH(thr_id, cuda_X[nxt], nxt, false);
cuda_scrypt_flush(thr_id, nxt);
if(!cuda_scrypt_sync(thr_id, cur))
{
return -1;
}
memcpy(Xbuf[cur].ptr, cuda_X[cur], 128 * throughput);
for(int i=0;i<throughput;++i)
scrypt_pbkdf2_1((unsigned char *)&data[cur][20*i], 80, Xbuf[cur].ptr + 128 * i, 128, (unsigned char *)(&hash[cur][8*i]), 32);
#define VERIFY_ALL 0
#if VERIFY_ALL
{
/* 2: X = ROMix(X) */
for(int i=0;i<throughput;++i)
scrypt_ROMix_1((scrypt_mix_word_t *)(Xbuf[cur].ptr + 128 * i), (scrypt_mix_word_t *)Ybuf.ptr, (scrypt_mix_word_t *)Vbuf.ptr, N);
unsigned int err = 0;
for(int i=0;i<throughput;++i) {
unsigned char *ref = (Xbuf[cur].ptr + 128 * i);
unsigned char *dat = (unsigned char*)(cuda_X[cur] + 32 * i);
if (memcmp(ref, dat, 128) != 0)
{
err++;
#if 0
uint32_t *ref32 = (uint32_t*) ref;
uint32_t *dat32 = (uint32_t*) dat;
for (int j=0; j<32; ++j) {
if (ref32[j] != dat32[j])
fprintf(stderr, "ref32[i=%d][j=%d] = $%08x / $%08x\n", i, j, ref32[j], dat32[j]);
}
#endif
}
}
if (err > 0) fprintf(stderr, "%d out of %d hashes differ.\n", err, throughput);
}
#endif
} else {
n += throughput;
cuda_scrypt_serialize(thr_id, nxt);
pre_keccak512(thr_id, nxt, nonce[nxt], throughput);
cuda_scrypt_core(thr_id, nxt, N);
cuda_scrypt_flush(thr_id, nxt);
post_keccak512(thr_id, nxt, nonce[nxt], throughput);
cuda_scrypt_done(thr_id, nxt);
cuda_scrypt_DtoH(thr_id, hash[nxt], nxt, true);
if(!cuda_scrypt_sync(thr_id, cur))
{
return -1;
}
}
if(iteration > 0)
{
for(int i=0;i<throughput;++i) {
volatile unsigned char *hashc = (unsigned char *)(&hash[cur][8*i]);
if (hash[cur][8*i+7] <= Htarg && fulltest(&hash[cur][8*i], ptarget)) {
uint32_t tmp_nonce = nonce[cur]+i;
uint32_t thash[8], tdata[20];
for(int z=0;z<20;z++) tdata[z] = bswap_32x4(pdata[z]);
tdata[19] = bswap_32x4(tmp_nonce);
scrypt_pbkdf2_1((unsigned char *)tdata, 80, (unsigned char *)tdata, 80, Xbuf[cur].ptr + 128 * i, 128);
scrypt_ROMix_1((scrypt_mix_word_t *)(Xbuf[cur].ptr + 128 * i), (scrypt_mix_word_t *)(Ybuf.ptr), (scrypt_mix_word_t *)(Vbuf.ptr), N);
scrypt_pbkdf2_1((unsigned char *)tdata, 80, Xbuf[cur].ptr + 128 * i, 128, (unsigned char *)thash, 32);
if (memcmp(thash, &hash[cur][8*i], 32) == 0)
{
//applog(LOG_INFO, "GPU #%d: %s result validates on CPU.", device_map[thr_id], device_name[thr_id]);
*hashes_done = n - pdata[19];
pdata[19] = tmp_nonce;
scrypt_free(&Vbuf);
scrypt_free(&Ybuf);
scrypt_free(&Xbuf[0]); scrypt_free(&Xbuf[1]);
delete[] data[0]; delete[] data[1];
gettimeofday(tv_end, NULL);
return 1;
}
else
{
applog(LOG_INFO, "GPU #%d: %s result does not validate on CPU (i=%d, s=%d)!", device_map[thr_id], device_name[thr_id], i, cur);
}
}
}
}
cur = (cur+1)&1;
nxt = (nxt+1)&1;
++iteration;
} while (n <= max_nonce && !work_restart[thr_id].restart);
scrypt_free(&Vbuf);
scrypt_free(&Ybuf);
scrypt_free(&Xbuf[0]); scrypt_free(&Xbuf[1]);
delete[] data[0]; delete[] data[1];
*hashes_done = n - pdata[19];
pdata[19] = n;
gettimeofday(tv_end, NULL);
return 0;
}
int scanhash_scrypt_jane(int thr_id, uint32_t *pdata,
const uint32_t *ptarget,
uint32_t max_nonce, unsigned long *hashes_done)
{
const uint32_t Htarg = ptarget[7];
int Nfactor = GetNfactor(bswap_32x4(pdata[17]));
if (Nfactor > scrypt_maxN) {
scrypt_fatal_error("scrypt: N out of range");
}
N = (1 << (Nfactor + 1));
parallel = 0;
int throughput = cuda_throughput(thr_id);
uint32_t *data[2] = { new uint32_t[20*throughput], new uint32_t[20*throughput] };
uint32_t *hash = new uint32_t[8*throughput];
uint32_t n = pdata[19] - 1;
// int i;
#if !defined(SCRYPT_TEST)
static int power_on_self_test = 0;
if (!power_on_self_test) {
power_on_self_test = 1;
if (!scrypt_power_on_self_test())
scrypt_fatal_error("scrypt: power on self test failed");
}
#endif
/* byte swap pdata into data[0]/[1] arrays */
for (int k=0; k<2; ++k) {
for(int z=0;z<20;z++) data[k][z] = bswap_32x4(pdata[z]);
for(int i=1;i<throughput;++i) memcpy(&data[k][20*i], &data[k][0], 20*sizeof(uint32_t));
}
scrypt_aligned_alloc Xbuf[2] = { scrypt_alloc(128 * throughput), scrypt_alloc(128 * throughput) };
scrypt_aligned_alloc Vbuf = scrypt_alloc((uint64_t)N * 128);
scrypt_aligned_alloc Ybuf = scrypt_alloc(128);
uint32_t nonce[2];
uint32_t* cuda_X[2] = { cuda_transferbuffer(thr_id,0), cuda_transferbuffer(thr_id,1) };
#if !defined(SCRYPT_CHOOSE_COMPILETIME)
scrypt_ROMixfn scrypt_ROMix = scrypt_getROMix();
#endif
int cur = 0, nxt = 1;
nonce[cur] = n+1;
for(int i=0;i<throughput;++i) {
uint32_t tmp_nonce = ++n;
data[cur][20*i + 19] = bswap_32x4(tmp_nonce);
}
/* 1: X = PBKDF2(password, salt) */
for(int i=0;i<throughput;++i)
scrypt_pbkdf2_1((unsigned char *)&data[cur][20*i], 80, (unsigned char *)&data[cur][20*i], 80, Xbuf[cur].ptr + 128 * i, 128);
/* 2: X = ROMix(X) in CUDA */
memcpy(cuda_X[cur], Xbuf[cur].ptr, 128 * throughput);
cuda_scrypt_HtoD(thr_id, cuda_X[cur], cur);
cuda_scrypt_serialize(thr_id, cur);
cuda_scrypt_core(thr_id, cur, N);
cuda_scrypt_done(thr_id, cur);
cuda_scrypt_DtoH(thr_id, cuda_X[cur], cur);
cuda_scrypt_flush(thr_id, cur);
do {
nonce[nxt] = n+1;
for(int i=0;i<throughput;++i) {
uint32_t tmp_nonce = ++n;
data[nxt][20*i + 19] = bswap_32x4(tmp_nonce);
}
/* 1: X = PBKDF2(password, salt) */
for(int i=0;i<throughput;++i)
scrypt_pbkdf2_1((unsigned char *)&data[nxt][20*i], 80, (unsigned char *)&data[nxt][20*i], 80, Xbuf[nxt].ptr + 128 * i, 128);
/* 2: X = ROMix(X) in CUDA */
memcpy(cuda_X[nxt], Xbuf[nxt].ptr, 128 * throughput);
cuda_scrypt_HtoD(thr_id, cuda_X[nxt], nxt);
cuda_scrypt_serialize(thr_id, nxt);
cuda_scrypt_core(thr_id, nxt, N);
cuda_scrypt_done(thr_id, nxt);
cuda_scrypt_DtoH(thr_id, cuda_X[nxt], nxt);
cuda_scrypt_flush(thr_id, nxt);
cuda_scrypt_sync(thr_id, cur);
#define VERIFY_ALL 0
#if VERIFY_ALL
{
/* 2: X = ROMix(X) */
for(int i=0;i<throughput;++i)
scrypt_ROMix_1((scrypt_mix_word_t *)(Xbuf[cur].ptr + 128 * i), (scrypt_mix_word_t *)Ybuf.ptr, (scrypt_mix_word_t *)Vbuf.ptr, N);
unsigned int err = 0;
for(int i=0;i<throughput;++i) {
unsigned char *ref = (Xbuf[cur].ptr + 128 * i);
unsigned char *dat = (unsigned char*)(cuda_X[cur] + 32 * i);
if (memcmp(ref, dat, 128) != 0)
{
err++;
#if 0
uint32_t *ref32 = (uint32_t*) ref;
uint32_t *dat32 = (uint32_t*) dat;
for (int j=0; j<32; ++j) {
if (ref32[j] != dat32[j])
fprintf(stderr, "ref32[i=%d][j=%d] = $%08x / $%08x\n", i, j, ref32[j], dat32[j]);
}
#endif
}
}
if (err > 0) fprintf(stderr, "%d out of %d hashes differ.\n", err, throughput);
}
#endif
memcpy(Xbuf[cur].ptr, cuda_X[cur], 128 * throughput);
/* 3: Out = PBKDF2(password, X) */
for(int i=0;i<throughput;++i)
scrypt_pbkdf2_1((unsigned char *)&data[cur][20*i], 80, Xbuf[cur].ptr + 128 * i, 128, (unsigned char *)&hash[8*i], 32);
for(int i=0;i<throughput;++i) {
volatile unsigned char *hashc = (unsigned char *)&hash[8*i];
if (hash[8*i+7] <= Htarg && fulltest(&hash[8*i], ptarget)) {
uint32_t tmp_nonce = nonce[cur]+i;
uint32_t thash[8], tdata[20];
for(int z=0;z<20;z++) tdata[z] = bswap_32x4(pdata[z]);
tdata[19] = bswap_32x4(tmp_nonce);
scrypt_pbkdf2_1((unsigned char *)tdata, 80, (unsigned char *)tdata, 80, Xbuf[cur].ptr + 128 * i, 128);
scrypt_ROMix_1((scrypt_mix_word_t *)(Xbuf[cur].ptr + 128 * i), (scrypt_mix_word_t *)(Ybuf.ptr), (scrypt_mix_word_t *)(Vbuf.ptr), N);
scrypt_pbkdf2_1((unsigned char *)tdata, 80, Xbuf[cur].ptr + 128 * i, 128, (unsigned char *)thash, 32);
if (memcmp(thash, &hash[8*i], 32) == 0)
{
*hashes_done = (n-throughput) - pdata[19] + 1;
pdata[19] = tmp_nonce;
scrypt_free(&Vbuf);
scrypt_free(&Ybuf);
scrypt_free(&Xbuf[0]); scrypt_free(&Xbuf[1]);
delete[] data[0]; delete[] data[1];
delete[] hash;
return 1;
}
else
{
applog(LOG_INFO, "GPU #%d: %s result does not validate on CPU (i=%d, s=%d)!", device_map[thr_id], device_name[thr_id], i, cur);
}
}
}
cur = (cur+1)&1; nxt = (nxt+1)&1;
} while ((n-throughput) < max_nonce && !work_restart[thr_id].restart);
scrypt_free(&Vbuf);
scrypt_free(&Ybuf);
scrypt_free(&Xbuf[0]); scrypt_free(&Xbuf[1]);
delete[] data[0]; delete[] data[1];
delete[] hash;
*hashes_done = (n-throughput) - pdata[19] + 1;
pdata[19] = (n-throughput);
return 0;
}
int scanhash_scryptjane( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done)
{
scrypt_aligned_alloc YX, V;
uint8_t *X, *Y;
uint32_t N, chunk_bytes;
const uint32_t r = SCRYPT_R;
const uint32_t p = SCRYPT_P;
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
uint32_t _ALIGN(64) endiandata[20];
const uint32_t first_nonce = pdata[19];
uint32_t nonce = first_nonce;
if (opt_benchmark)
ptarget[7] = 0x00ff;
for (int k = 0; k < 19; k++)
be32enc(&endiandata[k], pdata[k]);
//Nfactor = GetNfactor(data[17], ntime);
//if (Nfactor > scrypt_maxN) {
// return 1;
// //scrypt_fatal_error("scrypt: N out of range");
//}
N = (1 << ( opt_scrypt_n + 1));
chunk_bytes = SCRYPT_BLOCK_BYTES * r * 2;
if (!scrypt_alloc((uint64_t)N * chunk_bytes, &V)) return 1;
if (!scrypt_alloc((p + 1) * chunk_bytes, &YX)) {
scrypt_free(&V);
return 1;
}
Y = YX.ptr;
X = Y + chunk_bytes;
do {
const uint32_t Htarg = ptarget[7];
uint32_t hash[8];
be32enc(&endiandata[19], nonce);
scrypt_N_1_1((unsigned char *)endiandata, 80,
(unsigned char *)endiandata, 80,
N, (unsigned char *)hash, 32, X, Y, V.ptr);
if (hash[7] <= Htarg && fulltest(hash, ptarget)) {
pdata[19] = nonce;
*hashes_done = pdata[19] - first_nonce;
scrypt_free(&V);
scrypt_free(&YX);
return 1;
}
nonce++;
} while (nonce < max_nonce && !work_restart[thr_id].restart);
pdata[19] = nonce;
*hashes_done = pdata[19] - first_nonce + 1;
scrypt_free(&V);
scrypt_free(&YX);
return 0;
}
