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

v3.12.2

Browse Source
This commit is contained in:
Jay D Dee
2020-02-09 13:30:40 -05:00
parent dc2f8d81d3
commit 3da2b958cf
39 changed files with 1496 additions and 1518 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1244
algo/x17/sonoa-4way.c
View File

File diff suppressed because it is too large Load Diff

64
algo/x17/sonoa.c
View File

@@ -563,59 +563,31 @@ void sonoa_hash( void *state, const void *input )
}
int scanhash_sonoa( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
uint64_t *hashes_done, struct thr_info *mythr)
{
uint32_t _ALIGN(128) hash32[8];
uint32_t _ALIGN(128) endiandata[20];
uint32_t edata[20] __attribute__((aligned(64)));
uint32_t hash64[8] __attribute__((aligned(64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
uint32_t n = pdata[19];
const uint32_t first_nonce = pdata[19];
const uint32_t Htarg = ptarget[7];
uint32_t n = pdata[19] - 1;
int thr_id = mythr->id; // thr_id arg is deprecated
const int thr_id = mythr->id;
const bool bench = opt_benchmark;
uint64_t htmax[] =
mm128_bswap32_80( edata, pdata );
do
{
0,
0xF,
0xFF,
0xFFF,
0xFFFF,
0x10000000
};
uint32_t masks[] =
{
0xFFFFFFFF,
0xFFFFFFF0,
0xFFFFFF00,
0xFFFFF000,
0xFFFF0000,
0
};
// we need bigendian data...
casti_m128i( endiandata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
casti_m128i( endiandata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
casti_m128i( endiandata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
casti_m128i( endiandata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
casti_m128i( endiandata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
for ( int m = 0; m < 6; m++ ) if ( Htarg <= htmax[m] )
{
uint32_t mask = masks[m];
do
edata[19] = n;
sonoa_hash( hash64, edata );
if ( unlikely( valid_hash( hash64, ptarget ) && !bench ) )
{
pdata[19] = ++n;
be32enc(&endiandata[19], n);
sonoa_hash(hash32, endiandata);
if ( !( hash32[7] & mask ) )
if ( fulltest( hash32, ptarget ) && !opt_benchmark )
submit_solution( work, hash32, mythr );
} while (n < max_nonce && !work_restart[thr_id].restart);
break;
}
*hashes_done = n - first_nonce + 1;
pdata[19] = bswap_32( n );
submit_solution( work, hash64, mythr );
}
n++;
} while ( n < max_nonce && !work_restart[thr_id].restart );
*hashes_done = n - first_nonce;
pdata[19] = n;
return 0;
}

22
algo/x17/x17-4way.c
View File

@@ -74,9 +74,7 @@ void x17_8way_hash( void *state, const void *input )
blake512_8way_full( &ctx.blake, vhash, input, 80 );
bmw512_8way_init( &ctx.bmw );
bmw512_8way_update( &ctx.bmw, vhash, 64 );
bmw512_8way_close( &ctx.bmw, vhash );
bmw512_8way_full( &ctx.bmw, vhash, vhash, 64 );
#if defined(__VAES__)
@@ -106,9 +104,7 @@ void x17_8way_hash( void *state, const void *input )
#endif
skein512_8way_init( &ctx.skein );
skein512_8way_update( &ctx.skein, vhash, 64 );
skein512_8way_close( &ctx.skein, vhash );
skein512_8way_full( &ctx.skein, vhash, vhash, 64 );
jh512_8way_init( &ctx.jh );
jh512_8way_update( &ctx.jh, vhash, 64 );
@@ -290,12 +286,12 @@ int scanhash_x17_8way( struct work *work, uint32_t max_nonce,
uint32_t hash[8*8] __attribute__ ((aligned (128)));
uint32_t vdata[20*8] __attribute__ ((aligned (64)));
uint32_t lane_hash[8] __attribute__ ((aligned (64)));
uint32_t *hash32 = &(hash[7*8]);
uint32_t *hashd7 = &(hash[7*8]);
uint32_t *pdata = work->data;
const uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
const uint32_t last_nonce = max_nonce - 8;
__m512i *noncev = (__m512i*)vdata + 9; // aligned
__m512i *noncev = (__m512i*)vdata + 9;
uint32_t n = first_nonce;
const int thr_id = mythr->id;
const uint32_t targ32 = ptarget[7];
@@ -310,7 +306,7 @@ int scanhash_x17_8way( struct work *work, uint32_t max_nonce,
x17_8way_hash( hash, vdata );
for ( int lane = 0; lane < 8; lane++ )
if ( unlikely( ( hash32[ lane ] <= targ32 ) && !bench ) )
if ( unlikely( ( hashd7[ lane ] <= targ32 ) && !bench ) )
{
extr_lane_8x32( lane_hash, hash, lane, 256 );
if ( likely( valid_hash( lane_hash, ptarget ) ) )
@@ -378,9 +374,7 @@ void x17_4way_hash( void *state, const void *input )
intrlv_4x64_512( vhash, hash0, hash1, hash2, hash3 );
skein512_4way_init( &ctx.skein );
skein512_4way_update( &ctx.skein, vhash, 64 );
skein512_4way_close( &ctx.skein, vhash );
skein512_4way_full( &ctx.skein, vhash, vhash, 64 );
jh512_4way_init( &ctx.jh );
jh512_4way_update( &ctx.jh, vhash, 64 );
@@ -477,7 +471,7 @@ int scanhash_x17_4way( struct work *work, uint32_t max_nonce,
uint32_t hash[8*4] __attribute__ ((aligned (64)));
uint32_t vdata[20*4] __attribute__ ((aligned (64)));
uint32_t lane_hash[8] __attribute__ ((aligned (64)));
uint32_t *hash32 = &(hash[ 7*4 ]);
uint32_t *hashd7 = &(hash[ 7*4 ]);
uint32_t *pdata = work->data;
const uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
@@ -496,7 +490,7 @@ int scanhash_x17_4way( struct work *work, uint32_t max_nonce,
x17_4way_hash( hash, vdata );
for ( int lane = 0; lane < 4; lane++ )
if ( unlikely( hash32[ lane ] <= targ32 && !bench ) )
if ( unlikely( hashd7[ lane ] <= targ32 && !bench ) )
{
extr_lane_4x32( lane_hash, hash, lane, 256 );
if ( valid_hash( lane_hash, ptarget ) )

4
algo/x17/x17.c
View File

@@ -169,8 +169,8 @@ int scanhash_x17( struct work *work, uint32_t max_nonce,
submit_solution( work, hash64, mythr );
}
n++;
} while ( n < max_nonce && !work_restart[thr_id].restart);
*hashes_done = n - first_nonce + 1;
} while ( n < max_nonce && !work_restart[thr_id].restart );
*hashes_done = n - first_nonce;
pdata[19] = n;
return 0;
}

113
algo/x17/xevan-4way.c
View File

@@ -76,9 +76,7 @@ void xevan_8way_hash( void *output, const void *input )
blake512_8way_full( &ctx.blake, vhash, input, 80 );
memset( &vhash[8<<3], 0, 64<<3 );
bmw512_8way_init( &ctx.bmw );
bmw512_8way_update( &ctx.bmw, vhash, dataLen );
bmw512_8way_close( &ctx.bmw, vhash );
bmw512_8way_full( &ctx.bmw, vhash, vhash, dataLen );
#if defined(__VAES__)
@@ -108,9 +106,7 @@ void xevan_8way_hash( void *output, const void *input )
#endif
skein512_8way_init( &ctx.skein );
skein512_8way_update( &ctx.skein, vhash, dataLen );
skein512_8way_close( &ctx.skein, vhash );
skein512_8way_full( &ctx.skein, vhash, vhash, dataLen );
jh512_8way_init( &ctx.jh );
jh512_8way_update( &ctx.jh, vhash, dataLen );
@@ -291,9 +287,7 @@ void xevan_8way_hash( void *output, const void *input )
blake512_8way_full( &ctx.blake, vhash, vhash, dataLen );
bmw512_8way_init( &ctx.bmw );
bmw512_8way_update( &ctx.bmw, vhash, dataLen );
bmw512_8way_close( &ctx.bmw, vhash );
bmw512_8way_full( &ctx.bmw, vhash, vhash, dataLen );
#if defined(__VAES__)
@@ -323,9 +317,7 @@ void xevan_8way_hash( void *output, const void *input )
#endif
skein512_8way_init( &ctx.skein );
skein512_8way_update( &ctx.skein, vhash, dataLen );
skein512_8way_close( &ctx.skein, vhash );
skein512_8way_full( &ctx.skein, vhash, vhash, dataLen );
jh512_8way_init( &ctx.jh );
jh512_8way_update( &ctx.jh, vhash, dataLen );
@@ -504,40 +496,43 @@ void xevan_8way_hash( void *output, const void *input )
int scanhash_xevan_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t hash[8*16] __attribute__ ((aligned (128)));
uint32_t vdata[24*8] __attribute__ ((aligned (64)));
uint32_t hash[8*8] __attribute__ ((aligned (128)));
uint32_t vdata[20*8] __attribute__ ((aligned (64)));
uint32_t lane_hash[8] __attribute__ ((aligned (64)));
uint32_t *hash7 = &(hash[7<<3]);
uint32_t *hashd7 = &(hash[7*8]);
uint32_t *pdata = work->data;
const uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
const uint32_t last_nonce = max_nonce - 8;
__m512i *noncev = (__m512i*)vdata + 9; // aligned
__m512i *noncev = (__m512i*)vdata + 9;
uint32_t n = first_nonce;
const int thr_id = mythr->id;
const uint32_t Htarg = ptarget[7];
const uint32_t targ32 = ptarget[7];
const bool bench = opt_benchmark;
mm512_bswap32_intrlv80_8x64( vdata, pdata );
*noncev = mm512_intrlv_blend_32(
_mm512_set_epi32( n+7, 0, n+6, 0, n+5, 0, n+4, 0,
n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
do
{
*noncev = mm512_intrlv_blend_32( mm512_bswap_32(
_mm512_set_epi32( n+7, 0, n+6, 0, n+5, 0, n+4, 0,
n+3, 0, n+2, 0, n+1, 0, n, 0 ) ), *noncev );
xevan_8way_hash( hash, vdata );
for ( int lane = 0; lane < 8; lane++ )
if unlikely( ( hash7[ lane ] <= Htarg ) )
if ( unlikely( ( hashd7[ lane ] <= targ32 ) && !bench ) )
{
extr_lane_8x32( lane_hash, hash, lane, 256 );
if ( likely( fulltest( lane_hash, ptarget ) && !opt_benchmark ) )
if ( likely( valid_hash( lane_hash, ptarget ) ) )
{
pdata[19] = n + lane;
pdata[19] = bswap_32( n + lane );
submit_lane_solution( work, lane_hash, mythr, lane );
}
}
*noncev = _mm512_add_epi32( *noncev,
m512_const1_64( 0x0000000800000000 ) );
n += 8;
} while ( likely( ( n < last_nonce ) && !work_restart[thr_id].restart ) );
pdata[19] = n;
*hashes_done = n - first_nonce;
return 0;
}
@@ -578,8 +573,6 @@ void xevan_4way_hash( void *output, const void *input )
const int dataLen = 128;
xevan_4way_context_overlay ctx __attribute__ ((aligned (64)));
// parallel 4 way
blake512_4way_full( &ctx.blake, vhash, input, 80 );
memset( &vhash[8<<2], 0, 64<<2 );
@@ -598,9 +591,7 @@ void xevan_4way_hash( void *output, const void *input )
// Parallel 4way
intrlv_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
skein512_4way_init( &ctx.skein );
skein512_4way_update( &ctx.skein, vhash, dataLen );
skein512_4way_close( &ctx.skein, vhash );
skein512_4way_full( &ctx.skein, vhash, vhash, dataLen );
jh512_4way_init( &ctx.jh );
jh512_4way_update( &ctx.jh, vhash, dataLen );
@@ -618,15 +609,11 @@ void xevan_4way_hash( void *output, const void *input )
cube_2way_full( &ctx.cube, vhashA, 512, vhashA, dataLen );
cube_2way_full( &ctx.cube, vhashB, 512, vhashB, dataLen );
shavite512_2way_init( &ctx.shavite );
shavite512_2way_update_close( &ctx.shavite, vhashA, vhashA, dataLen );
shavite512_2way_init( &ctx.shavite );
shavite512_2way_update_close( &ctx.shavite, vhashB, vhashB, dataLen );
shavite512_2way_full( &ctx.shavite, vhashA, vhashA, dataLen );
shavite512_2way_full( &ctx.shavite, vhashB, vhashB, dataLen );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhashA, vhashA, dataLen<<3 );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhashB, vhashB, dataLen<<3 );
simd512_2way_full( &ctx.simd, vhashA, vhashA, dataLen );
simd512_2way_full( &ctx.simd, vhashB, vhashB, dataLen );
dintrlv_2x128( hash0, hash1, vhashA, dataLen<<3 );
dintrlv_2x128( hash2, hash3, vhashB, dataLen<<3 );
@@ -718,9 +705,7 @@ void xevan_4way_hash( void *output, const void *input )
intrlv_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
skein512_4way_init( &ctx.skein );
skein512_4way_update( &ctx.skein, vhash, dataLen );
skein512_4way_close( &ctx.skein, vhash );
skein512_4way_full( &ctx.skein, vhash, vhash, dataLen );
jh512_4way_init( &ctx.jh );
jh512_4way_update( &ctx.jh, vhash, dataLen );
@@ -738,15 +723,11 @@ void xevan_4way_hash( void *output, const void *input )
cube_2way_full( &ctx.cube, vhashA, 512, vhashA, dataLen );
cube_2way_full( &ctx.cube, vhashB, 512, vhashB, dataLen );
shavite512_2way_init( &ctx.shavite );
shavite512_2way_update_close( &ctx.shavite, vhashA, vhashA, dataLen );
shavite512_2way_init( &ctx.shavite );
shavite512_2way_update_close( &ctx.shavite, vhashB, vhashB, dataLen );
shavite512_2way_full( &ctx.shavite, vhashA, vhashA, dataLen );
shavite512_2way_full( &ctx.shavite, vhashB, vhashB, dataLen );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhashA, vhashA, dataLen<<3 );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhashB, vhashB, dataLen<<3 );
simd512_2way_full( &ctx.simd, vhashA, vhashA, dataLen );
simd512_2way_full( &ctx.simd, vhashB, vhashB, dataLen );
dintrlv_2x128( hash0, hash1, vhashA, dataLen<<3 );
dintrlv_2x128( hash2, hash3, vhashB, dataLen<<3 );
@@ -818,41 +799,43 @@ void xevan_4way_hash( void *output, const void *input )
int scanhash_xevan_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t hash[4*16] __attribute__ ((aligned (64)));
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t lane_hash[8] __attribute__ ((aligned (32)));
uint32_t *hash7 = &(hash[7<<2]);
uint32_t hash[16*4] __attribute__ ((aligned (128)));
uint32_t vdata[20*4] __attribute__ ((aligned (64)));
uint32_t lane_hash[8] __attribute__ ((aligned (64)));
uint32_t *hashd7 = &(hash[7<<2]);
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
int thr_id = mythr->id;
__m256i *noncev = (__m256i*)vdata + 9; // aligned
const uint32_t Htarg = ptarget[7];
__m256i *noncev = (__m256i*)vdata + 9;
const uint32_t targ32 = ptarget[7];
const uint32_t first_nonce = pdata[19];
const uint32_t last_nonce = max_nonce - 4;
uint32_t n = first_nonce;
const bool bench = opt_benchmark;
if ( opt_benchmark )
ptarget[7] = 0x0cff;
if ( bench ) ptarget[7] = 0x0cff;
mm256_bswap32_intrlv80_4x64( vdata, pdata );
*noncev = mm256_intrlv_blend_32(
_mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
do {
*noncev = mm256_intrlv_blend_32( mm256_bswap_32(
_mm256_set_epi32( n+3, 0,n+2, 0,n+1, 0, n, 0 ) ), *noncev );
xevan_4way_hash( hash, vdata );
for ( int lane = 0; lane < 4; lane++ )
if ( hash7[ lane ] <= Htarg )
if ( unlikely( hashd7[ lane ] <= targ32 ) && ! bench )
{
extr_lane_4x32( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
if ( valid_hash( lane_hash, ptarget ) )
{
pdata[19] = n + lane;
pdata[19] = bswap_32( n + lane );
submit_lane_solution( work, lane_hash, mythr, lane );
}
}
*noncev = _mm256_add_epi32( *noncev,
m256_const1_64( 0x0000000400000000 ) );
n += 4;
} while ( ( n < max_nonce-4 ) && !work_restart[thr_id].restart );
*hashes_done = n - first_nonce + 1;
} while ( likely( ( n < last_nonce ) && !work_restart[thr_id].restart ) );
pdata[19] = n;
*hashes_done = n - first_nonce;
return 0;
}

128
algo/x17/xevan.c
View File

@@ -56,8 +56,6 @@ typedef struct {
} xevan_ctx_holder;
xevan_ctx_holder xevan_ctx __attribute__ ((aligned (64)));
static __thread sph_blake512_context xevan_blake_mid
__attribute__ ((aligned (64)));
void init_xevan_ctx()
{
@@ -85,34 +83,23 @@ void init_xevan_ctx()
#endif
};
void xevan_blake512_midstate( const void* input )
{
memcpy( &xevan_blake_mid, &xevan_ctx.blake, sizeof xevan_blake_mid );
sph_blake512( &xevan_blake_mid, input, 64 );
}
void xevan_hash(void *output, const void *input)
{
uint32_t _ALIGN(64) hash[32]; // 128 bytes required
uint32_t _ALIGN(64) hash[32]; // 128 bytes required
const int dataLen = 128;
xevan_ctx_holder ctx __attribute__ ((aligned (64)));
memcpy( &ctx, &xevan_ctx, sizeof(xevan_ctx) );
const int midlen = 64; // bytes
const int tail = 80 - midlen; // 16
memcpy( &ctx.blake, &xevan_blake_mid, sizeof xevan_blake_mid );
sph_blake512( &ctx.blake, input + midlen, tail );
sph_blake512_close(&ctx.blake, hash);
xevan_ctx_holder ctx __attribute__ ((aligned (64)));
memcpy( &ctx, &xevan_ctx, sizeof(xevan_ctx) );
sph_blake512( &ctx.blake, input, 80 );
sph_blake512_close( &ctx.blake, hash );
memset(&hash[16], 0, 64);
sph_bmw512(&ctx.bmw, hash, dataLen);
sph_bmw512_close(&ctx.bmw, hash);
#if defined(__AES__)
update_and_final_groestl( &ctx.groestl, (char*)hash,
(const char*)hash, dataLen*8 );
update_and_final_groestl( &ctx.groestl, (char*)hash,
(const char*)hash, dataLen*8 );
#else
sph_groestl512(&ctx.groestl, hash, dataLen);
sph_groestl512_close(&ctx.groestl, hash);
@@ -127,20 +114,20 @@ void xevan_hash(void *output, const void *input)
sph_keccak512(&ctx.keccak, hash, dataLen);
sph_keccak512_close(&ctx.keccak, hash);
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash,
(const BitSequence*)hash, dataLen );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash,
(const BitSequence*)hash, dataLen );
cubehashUpdateDigest( &ctx.cubehash, (byte*)hash,
(const byte*) hash, dataLen );
cubehashUpdateDigest( &ctx.cubehash, (byte*)hash,
(const byte*) hash, dataLen );
sph_shavite512(&ctx.shavite, hash, dataLen);
sph_shavite512_close(&ctx.shavite, hash);
update_final_sd( &ctx.simd, (BitSequence *)hash,
update_final_sd( &ctx.simd, (BitSequence *)hash,
(const BitSequence *)hash, dataLen*8 );
#if defined(__AES__)
update_final_echo( &ctx.echo, (BitSequence *) hash,
update_final_echo( &ctx.echo, (BitSequence *) hash,
(const BitSequence *) hash, dataLen*8 );
#else
sph_echo512(&ctx.echo, hash, dataLen);
@@ -159,15 +146,15 @@ void xevan_hash(void *output, const void *input)
sph_whirlpool(&ctx.whirlpool, hash, dataLen);
sph_whirlpool_close(&ctx.whirlpool, hash);
SHA512_Update( &ctx.sha512, hash, dataLen );
SHA512_Final( (unsigned char*) hash, &ctx.sha512 );
SHA512_Update( &ctx.sha512, hash, dataLen );
SHA512_Final( (unsigned char*) hash, &ctx.sha512 );
sph_haval256_5(&ctx.haval,(const void*) hash, dataLen);
sph_haval256_5_close(&ctx.haval, hash);
memset(&hash[8], 0, dataLen - 32);
memcpy( &ctx, &xevan_ctx, sizeof(xevan_ctx) );
memcpy( &ctx, &xevan_ctx, sizeof(xevan_ctx) );
sph_blake512(&ctx.blake, hash, dataLen);
sph_blake512_close(&ctx.blake, hash);
@@ -176,11 +163,11 @@ void xevan_hash(void *output, const void *input)
sph_bmw512_close(&ctx.bmw, hash);
#if defined(__AES__)
update_and_final_groestl( &ctx.groestl, (char*)hash,
(const BitSequence*)hash, dataLen*8 );
update_and_final_groestl( &ctx.groestl, (char*)hash,
(const BitSequence*)hash, dataLen*8 );
#else
sph_groestl512(&ctx.groestl, hash, dataLen);
sph_groestl512_close(&ctx.groestl, hash);
sph_groestl512_close(&ctx.groestl, hash);
#endif
sph_skein512(&ctx.skein, hash, dataLen);
@@ -191,24 +178,25 @@ void xevan_hash(void *output, const void *input)
sph_keccak512(&ctx.keccak, hash, dataLen);
sph_keccak512_close(&ctx.keccak, hash);
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash,
(const BitSequence*)hash, dataLen );
cubehashUpdateDigest( &ctx.cubehash, (byte*)hash,
(const byte*) hash, dataLen );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash,
(const BitSequence*)hash, dataLen );
cubehashUpdateDigest( &ctx.cubehash, (byte*)hash,
(const byte*) hash, dataLen );
sph_shavite512(&ctx.shavite, hash, dataLen);
sph_shavite512_close(&ctx.shavite, hash);
update_final_sd( &ctx.simd, (BitSequence *)hash,
update_final_sd( &ctx.simd, (BitSequence *)hash,
(const BitSequence *)hash, dataLen*8 );
#if defined(__AES__)
update_final_echo( &ctx.echo, (BitSequence *) hash,
update_final_echo( &ctx.echo, (BitSequence *) hash,
(const BitSequence *) hash, dataLen*8 );
#else
sph_echo512(&ctx.echo, hash, dataLen);
sph_echo512_close(&ctx.echo, hash);
sph_echo512(&ctx.echo, hash, dataLen);
sph_echo512_close(&ctx.echo, hash);
#endif
sph_hamsi512(&ctx.hamsi, hash, dataLen);
@@ -223,8 +211,8 @@ void xevan_hash(void *output, const void *input)
sph_whirlpool(&ctx.whirlpool, hash, dataLen);
sph_whirlpool_close(&ctx.whirlpool, hash);
SHA512_Update( &ctx.sha512, hash, dataLen );
SHA512_Final( (unsigned char*) hash, &ctx.sha512 );
SHA512_Update( &ctx.sha512, hash, dataLen );
SHA512_Final( (unsigned char*) hash, &ctx.sha512 );
sph_haval256_5(&ctx.haval,(const void*) hash, dataLen);
sph_haval256_5_close(&ctx.haval, hash);
@@ -233,41 +221,33 @@ void xevan_hash(void *output, const void *input)
}
int scanhash_xevan( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
uint64_t *hashes_done, struct thr_info *mythr)
{
uint32_t _ALIGN(64) hash[8];
uint32_t _ALIGN(64) endiandata[20];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
int thr_id = mythr->id; // thr_id arg is deprecated
const uint32_t Htarg = ptarget[7];
const uint32_t first_nonce = pdata[19];
uint32_t nonce = first_nonce;
volatile uint8_t *restart = &(work_restart[thr_id].restart);
uint32_t edata[20] __attribute__((aligned(64)));
uint32_t hash64[8] __attribute__((aligned(64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
uint32_t n = pdata[19];
const uint32_t first_nonce = pdata[19];
const int thr_id = mythr->id;
const bool bench = opt_benchmark;
if (opt_benchmark)
ptarget[7] = 0x0cff;
mm128_bswap32_80( edata, pdata );
for (int k=0; k < 19; k++)
be32enc(&endiandata[k], pdata[k]);
xevan_blake512_midstate( endiandata );
do {
be32enc(&endiandata[19], nonce);
xevan_hash(hash, endiandata);
if (hash[7] <= Htarg )
if ( fulltest( hash, ptarget ) && !opt_benchmark )
{
pdata[19] = nonce;
submit_solution( work, hash, mythr );
}
nonce++;
} while ( nonce < max_nonce && !(*restart) );
pdata[19] = nonce;
*hashes_done = pdata[19] - first_nonce + 1;
return 0;
do
{
edata[19] = n;
xevan_hash( hash64, edata );
if ( unlikely( valid_hash( hash64, ptarget ) && !bench ) )
{
pdata[19] = bswap_32( n );
submit_solution( work, hash64, mythr );
}
n++;
} while ( n < max_nonce && !work_restart[thr_id].restart );
pdata[19] = n;
*hashes_done = n - first_nonce;
return 0;
}
#endif