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

v3.10.5

Browse Source
This commit is contained in:
Jay D Dee
2019-12-21 13:19:29 -05:00
parent a17ff6f189
commit c65b0ff7a6
72 changed files with 9090 additions and 1336 deletions
Download Patch File Download Diff File Expand all files Collapse all files

282
algo/x12/x12-4way.c
View File

@@ -1,7 +1,4 @@
#include "x12-gate.h"
#if defined(X12_4WAY)
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
@@ -14,11 +11,223 @@
#include "algo/keccak/keccak-hash-4way.h"
#include "algo/luffa/luffa-hash-2way.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/cubehash/cube-hash-2way.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/simd/simd-hash-2way.h"
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/hamsi/hamsi-hash-4way.h"
//#include "algo/fugue/sph_fugue.h"
#if defined(X12_8WAY)
typedef struct {
blake512_8way_context blake;
bmw512_8way_context bmw;
hashState_groestl groestl;
skein512_8way_context skein;
jh512_8way_context jh;
keccak512_8way_context keccak;
luffa_4way_context luffa;
cube_4way_context cube;
sph_shavite512_context shavite;
simd_4way_context simd;
hashState_echo echo;
hamsi512_8way_context hamsi;
} x12_8way_ctx_holder;
x12_8way_ctx_holder x12_8way_ctx __attribute__ ((aligned (64)));
void init_x12_8way_ctx()
{
blake512_8way_init( &x12_8way_ctx.blake );
bmw512_8way_init( &x12_8way_ctx.bmw );
init_groestl( &x12_8way_ctx.groestl, 64 );
skein512_8way_init( &x12_8way_ctx.skein );
jh512_8way_init( &x12_8way_ctx.jh );
keccak512_8way_init( &x12_8way_ctx.keccak );
luffa_4way_init( &x12_8way_ctx.luffa, 512 );
cube_4way_init( &x12_8way_ctx.cube, 512, 16, 32 );
sph_shavite512_init( &x12_8way_ctx.shavite );
simd_4way_init( &x12_8way_ctx.simd, 512 );
init_echo( &x12_8way_ctx.echo, 512 );
hamsi512_8way_init( &x12_8way_ctx.hamsi );
};
void x12_8way_hash( void *state, const void *input )
{
uint64_t vhash[8*8] __attribute__ ((aligned (128)));
uint64_t vhash0[4*8] __attribute__ ((aligned (64)));
uint64_t vhash1[4*8] __attribute__ ((aligned (64)));
uint64_t hash0[8] __attribute__ ((aligned (64)));
uint64_t hash1[8] __attribute__ ((aligned (64)));
uint64_t hash2[8] __attribute__ ((aligned (64)));
uint64_t hash3[8] __attribute__ ((aligned (64)));
uint64_t hash4[8] __attribute__ ((aligned (64)));
uint64_t hash5[8] __attribute__ ((aligned (64)));
uint64_t hash6[8] __attribute__ ((aligned (64)));
uint64_t hash7[8] __attribute__ ((aligned (64)));
x12_8way_ctx_holder ctx;
memcpy( &ctx, &x12_8way_ctx, sizeof(x12_8way_ctx) );
blake512_8way_update( &ctx.blake, input, 80 );
blake512_8way_close( &ctx.blake, vhash );
bmw512_8way_update( &ctx.bmw, vhash, 64 );
bmw512_8way_close( &ctx.bmw, vhash );
rintrlv_8x64_4x128( vhash0, vhash1, vhash, 512 );
luffa_4way_update_close( &ctx.luffa, vhash0, vhash0, 64 );
luffa_4way_init( &ctx.luffa, 512 );
luffa_4way_update_close( &ctx.luffa, vhash1, vhash1, 64 );
cube_4way_update_close( &ctx.cube, vhash0, vhash0, 64 );
cube_4way_init( &ctx.cube, 512, 16, 32 );
cube_4way_update_close( &ctx.cube, vhash1, vhash1, 64 );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash0 );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash1 );
sph_shavite512( &ctx.shavite, hash0, 64 );
sph_shavite512_close( &ctx.shavite, hash0 );
memcpy( &ctx.shavite, &x12_8way_ctx.shavite,
sizeof(sph_shavite512_context) );
sph_shavite512( &ctx.shavite, hash1, 64 );
sph_shavite512_close( &ctx.shavite, hash1 );
memcpy( &ctx.shavite, &x12_8way_ctx.shavite,
sizeof(sph_shavite512_context) );
sph_shavite512( &ctx.shavite, hash2, 64 );
sph_shavite512_close( &ctx.shavite, hash2 );
memcpy( &ctx.shavite, &x12_8way_ctx.shavite,
sizeof(sph_shavite512_context) );
sph_shavite512( &ctx.shavite, hash3, 64 );
sph_shavite512_close( &ctx.shavite, hash3 );
memcpy( &ctx.shavite, &x12_8way_ctx.shavite,
sizeof(sph_shavite512_context) );
sph_shavite512( &ctx.shavite, hash4, 64 );
sph_shavite512_close( &ctx.shavite, hash4 );
memcpy( &ctx.shavite, &x12_8way_ctx.shavite,
sizeof(sph_shavite512_context) );
sph_shavite512( &ctx.shavite, hash5, 64 );
sph_shavite512_close( &ctx.shavite, hash5 );
memcpy( &ctx.shavite, &x12_8way_ctx.shavite,
sizeof(sph_shavite512_context) );
sph_shavite512( &ctx.shavite, hash6, 64 );
sph_shavite512_close( &ctx.shavite, hash6 );
memcpy( &ctx.shavite, &x12_8way_ctx.shavite,
sizeof(sph_shavite512_context) );
sph_shavite512( &ctx.shavite, hash7, 64 );
sph_shavite512_close( &ctx.shavite, hash7 );
intrlv_4x128_512( vhash, hash0, hash1, hash2, hash3 );
simd_4way_update_close( &ctx.simd, vhash, vhash, 512 );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128_512( vhash, hash4, hash5, hash6, hash7 );
simd_4way_init( &ctx.simd, 512 );
simd_4way_update_close( &ctx.simd, vhash, vhash, 512 );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
update_final_echo( &ctx.echo, (BitSequence *)hash1,
(const BitSequence *) hash1, 512 );
memcpy( &ctx.echo, &x12_8way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash2,
(const BitSequence *) hash2, 512 );
memcpy( &ctx.echo, &x12_8way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash3,
(const BitSequence *) hash3, 512 );
memcpy( &ctx.echo, &x12_8way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash4,
(const BitSequence *) hash4, 512 );
memcpy( &ctx.echo, &x12_8way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash5,
(const BitSequence *) hash5, 512 );
memcpy( &ctx.echo, &x12_8way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash6,
(const BitSequence *) hash6, 512 );
memcpy( &ctx.echo, &x12_8way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash7,
(const BitSequence *) hash7, 512 );
update_and_final_groestl( &ctx.groestl, (char*)hash0, (char*)hash0, 512 );
memcpy( &ctx.groestl, &x12_8way_ctx.groestl,
sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash1, (char*)hash1, 512 );
memcpy( &ctx.groestl, &x12_8way_ctx.groestl,
sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash2, (char*)hash2, 512 );
memcpy( &ctx.groestl, &x12_8way_ctx.groestl,
sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash3, (char*)hash3, 512 );
memcpy( &ctx.groestl, &x12_8way_ctx.groestl,
sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash4, (char*)hash4, 512 );
memcpy( &ctx.groestl, &x12_8way_ctx.groestl,
sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash5, (char*)hash5, 512 );
memcpy( &ctx.groestl, &x12_8way_ctx.groestl,
sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash6, (char*)hash6, 512 );
memcpy( &ctx.groestl, &x12_8way_ctx.groestl,
sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash7, (char*)hash7, 512 );
intrlv_8x64_512( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7 );
skein512_8way_update( &ctx.skein, vhash, 64 );
skein512_8way_close( &ctx.skein, vhash );
jh512_8way_update( &ctx.jh, vhash, 64 );
jh512_8way_close( &ctx.jh, vhash );
keccak512_8way_update( &ctx.keccak, vhash, 64 );
keccak512_8way_close( &ctx.keccak, vhash );
hamsi512_8way_update( &ctx.hamsi, vhash, 64 );
hamsi512_8way_close( &ctx.hamsi, state );
}
int scanhash_x12_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t hash[16*8] __attribute__ ((aligned (128)));
uint32_t vdata[24*8] __attribute__ ((aligned (64)));
uint32_t lane_hash[8] __attribute__ ((aligned (64)));
uint32_t *hash7 = &(hash[49]);
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];
__m512i *noncev = (__m512i*)vdata + 9; // aligned
int thr_id = mythr->id;
mm512_bswap32_intrlv80_8x64( vdata, pdata );
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 );
x12_8way_hash( hash, vdata );
for ( int lane = 0; lane < 8; lane++ )
if ( hash7[ lane<<1 ] < Htarg )
{
extr_lane_8x64( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
{
pdata[19] = n + lane;
submit_lane_solution( work, lane_hash, mythr, lane );
}
}
n += 8;
} while ( ( n < max_nonce-8 ) && !work_restart[thr_id].restart );
*hashes_done = n - first_nonce;
return 0;
}
#elif defined(X12_4WAY)
typedef struct {
blake512_4way_context blake;
@@ -63,45 +272,13 @@ void x12_4way_hash( void *state, const void *input )
x12_4way_ctx_holder ctx;
memcpy( &ctx, &x12_4way_ctx, sizeof(x12_4way_ctx) );
// 1 Blake
blake512_4way( &ctx.blake, input, 80 );
blake512_4way_close( &ctx.blake, vhash );
// 2 Bmw
bmw512_4way( &ctx.bmw, vhash, 64 );
bmw512_4way_close( &ctx.bmw, vhash );
// Serial
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// 3 Groestl
update_and_final_groestl( &ctx.groestl, (char*)hash0, (char*)hash0, 512 );
memcpy( &ctx.groestl, &x12_4way_ctx.groestl, sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash1, (char*)hash1, 512 );
memcpy( &ctx.groestl, &x12_4way_ctx.groestl, sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash2, (char*)hash2, 512 );
memcpy( &ctx.groestl, &x12_4way_ctx.groestl, sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash3, (char*)hash3, 512 );
// Parallel 4way 64 bit
intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
// 4 Skein
skein512_4way( &ctx.skein, vhash, 64 );
skein512_4way_close( &ctx.skein, vhash );
// 5 JH
jh512_4way( &ctx.jh, vhash, 64 );
jh512_4way_close( &ctx.jh, vhash );
// 6 Keccak
keccak512_4way( &ctx.keccak, vhash, 64 );
keccak512_4way_close( &ctx.keccak, vhash );
// Serial
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// 7 Luffa
intrlv_2x128( vhash, hash0, hash1, 512 );
luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
dintrlv_2x128( hash0, hash1, vhash, 512 );
@@ -110,7 +287,6 @@ void x12_4way_hash( void *state, const void *input )
luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
dintrlv_2x128( hash2, hash3, vhash, 512 );
// 8 Cubehash
cubehashUpdateDigest( &ctx.cube, (byte*)hash0, (const byte*) hash0, 64 );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash1, (const byte*) hash1, 64 );
@@ -119,7 +295,6 @@ void x12_4way_hash( void *state, const void *input )
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash3, (const byte*) hash3, 64 );
// 9 Shavite
sph_shavite512( &ctx.shavite, hash0, 64 );
sph_shavite512_close( &ctx.shavite, hash0 );
memcpy( &ctx.shavite, &x12_4way_ctx.shavite,
@@ -135,7 +310,6 @@ void x12_4way_hash( void *state, const void *input )
sph_shavite512( &ctx.shavite, hash3, 64 );
sph_shavite512_close( &ctx.shavite, hash3 );
// 10 Simd
intrlv_2x128( vhash, hash0, hash1, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, 512 );
dintrlv_2x128( hash0, hash1, vhash, 512 );
@@ -144,21 +318,25 @@ void x12_4way_hash( void *state, const void *input )
simd_2way_update_close( &ctx.simd, vhash, vhash, 512 );
dintrlv_2x128( hash2, hash3, vhash, 512 );
// 11 Echo
update_final_echo( &ctx.echo, (BitSequence *)hash0,
(const BitSequence *) hash0, 512 );
memcpy( &ctx.echo, &x12_4way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash1,
(const BitSequence *) hash1, 512 );
memcpy( &ctx.echo, &x12_4way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash2,
(const BitSequence *) hash2, 512 );
memcpy( &ctx.echo, &x12_4way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash3,
(const BitSequence *) hash3, 512 );
update_and_final_groestl( &ctx.groestl, (char*)hash0, (char*)hash0, 512 );
memcpy( &ctx.groestl, &x12_4way_ctx.groestl, sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash1, (char*)hash1, 512 );
memcpy( &ctx.groestl, &x12_4way_ctx.groestl, sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash2, (char*)hash2, 512 );
memcpy( &ctx.groestl, &x12_4way_ctx.groestl, sizeof(hashState_groestl) );
update_and_final_groestl( &ctx.groestl, (char*)hash3, (char*)hash3, 512 );
// 12 Hamsi parallel 4way 32 bit
// Parallel 4way 64 bit
intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
skein512_4way( &ctx.skein, vhash, 64 );
skein512_4way_close( &ctx.skein, vhash );
jh512_4way( &ctx.jh, vhash, 64 );
jh512_4way_close( &ctx.jh, vhash );
keccak512_4way( &ctx.keccak, vhash, 64 );
keccak512_4way_close( &ctx.keccak, vhash );
hamsi512_4way( &ctx.hamsi, vhash, 64 );
hamsi512_4way_close( &ctx.hamsi, vhash );

8
algo/x12/x12-gate.c
View File

@@ -2,7 +2,11 @@
bool register_x12_algo( algo_gate_t* gate )
{
#if defined (X12_4WAY)
#if defined (X12_8WAY)
init_x12_8way_ctx();
gate->scanhash = (void*)&scanhash_x12_8way;
gate->hash = (void*)&x12_8way_hash;
#elif defined (X12_4WAY)
init_x12_4way_ctx();
gate->scanhash = (void*)&scanhash_x12_4way;
gate->hash = (void*)&x12_4way_hash;
@@ -11,7 +15,7 @@ bool register_x12_algo( algo_gate_t* gate )
gate->scanhash = (void*)&scanhash_x12;
gate->hash = (void*)&x12hash;
#endif
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT;
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT;
return true;
};

23
algo/x12/x12-gate.h
View File

@@ -4,29 +4,36 @@
#include "algo-gate-api.h"
#include <stdint.h>
#if defined(__AVX2__) && defined(__AES__)
#define X12_4WAY
#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
#define X12_8WAY 1
#elif defined(__AVX2__) && defined(__AES__)
#define X12_4WAY 1
#endif
bool register_x12_algo( algo_gate_t* gate );
#if defined(X12_4WAY)
#if defined(X12_8WAY)
void x12_8way_hash( void *state, const void *input );
int scanhash_x12_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
void init_x12_8way_ctx();
#elif defined(X12_4WAY)
void x12_4way_hash( void *state, const void *input );
int scanhash_x12_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
void init_x12_4way_ctx();
#endif
#else
void x12hash( void *state, const void *input );
int scanhash_x12( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
void init_x12_ctx();
#endif
#endif

146
algo/x12/x12.c
View File

@@ -20,35 +20,40 @@
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/nist.h"
#include "algo/blake/sse2/blake.c"
#include "algo/bmw/sse2/bmw.c"
#include "algo/keccak/sse2/keccak.c"
#include "algo/skein/sse2/skein.c"
#include "algo/jh/sse2/jh_sse2_opt64.h"
#if defined(__AES__)
#include "algo/groestl/aes_ni/hash-groestl.h"
#include "algo/echo/aes_ni/hash_api.h"
#endif
typedef struct {
sph_blake512_context blake;
sph_bmw512_context bmw;
sph_skein512_context skein;
sph_jh512_context jh;
sph_keccak512_context keccak;
#if defined(__AES__)
hashState_groestl groestl;
hashState_echo echo;
hashState_groestl groestl;
hashState_echo echo;
#else
sph_groestl512_context groestl;
sph_echo512_context echo;
sph_groestl512_context groestl;
sph_echo512_context echo;
#endif
hashState_luffa luffa;
cubehashParam cubehash;
sph_shavite512_context shavite;
hashState_sd simd;
sph_hamsi512_context hamsi;
hashState_luffa luffa;
cubehashParam cubehash;
sph_shavite512_context shavite;
hashState_sd simd;
sph_hamsi512_context hamsi;
} x12_ctx_holder;
x12_ctx_holder x12_ctx;
void init_x12_ctx()
{
sph_blake512_init( &x12_ctx.blake );
sph_bmw512_init( &x12_ctx.bmw );
sph_skein512_init( &x12_ctx.skein);
sph_jh512_init( &x12_ctx.jh);
sph_keccak512_init( &x12_ctx.keccak);
#if defined(__AES__)
init_echo( &x12_ctx.echo, 512 );
init_groestl (&x12_ctx.groestl, 64 );
@@ -65,102 +70,59 @@ void init_x12_ctx()
void x12hash(void *output, const void *input)
{
unsigned char hash[128] __attribute__ ((aligned (32)));
#define hashB hash+64
x12_ctx_holder ctx;
memcpy( &ctx, &x12_ctx, sizeof(x12_ctx) );
x12_ctx_holder ctx;
memcpy( &ctx, &x12_ctx, sizeof(x12_ctx) );
// X11 algos
sph_blake512(&ctx.blake, input, 80);
sph_blake512_close(&ctx.blake, hash);
unsigned char hashbuf[128];
size_t hashptr;
sph_u64 hashctA;
sph_u64 hashctB;
sph_bmw512(&ctx.bmw, hash, 64);
sph_bmw512_close(&ctx.bmw, hash);
//---blake1---
update_and_final_luffa( &ctx.luffa, (BitSequence*)hashB,
(const BitSequence*)hash, 64 );
DECL_BLK;
BLK_I;
BLK_W;
BLK_C;
cubehashUpdateDigest( &ctx.cubehash, (byte*) hash,
(const byte*)hashB, 64 );
//---bmw2---
sph_shavite512( &ctx.shavite, hash, 64);
sph_shavite512_close( &ctx.shavite, hashB);
DECL_BMW;
BMW_I;
BMW_U;
#define M(x) sph_dec64le_aligned(data + 8 * (x))
#define H(x) (h[x])
#define dH(x) (dh[x])
BMW_C;
#undef M
#undef H
#undef dH
//---groetl----
update_final_sd( &ctx.simd, (BitSequence *)hash,
(const BitSequence *)hashB, 512 );
#if defined(__AES__)
update_and_final_groestl( &ctx.groestl, (char*)hash,
(const char*)hash, 512 );
#else
sph_groestl512 (&ctx.groestl, hash, 64);
sph_groestl512_close(&ctx.groestl, hash);
#endif
//---skein4---
DECL_SKN;
SKN_I;
SKN_U;
SKN_C;
//---jh5------
DECL_JH;
JH_H;
//---keccak6---
DECL_KEC;
KEC_I;
KEC_U;
KEC_C;
//--- luffa7
update_and_final_luffa( &ctx.luffa, (BitSequence*)hashB,
(const BitSequence*)hash, 64 );
// 8 Cube
cubehashUpdateDigest( &ctx.cubehash, (byte*) hash,
(const byte*)hashB, 64 );
// 9 Shavite
sph_shavite512( &ctx.shavite, hash, 64);
sph_shavite512_close( &ctx.shavite, hashB);
// 10 Simd
update_final_sd( &ctx.simd, (BitSequence *)hash,
(const BitSequence *)hashB, 512 );
//11---echo---
#if defined(__AES__)
update_final_echo ( &ctx.echo, (BitSequence *)hashB,
update_final_echo ( &ctx.echo, (BitSequence *)hashB,
(const BitSequence *)hash, 512 );
#else
sph_echo512(&ctx.echo, hash, 64);
sph_echo512_close(&ctx.echo, hashB);
sph_echo512(&ctx.echo, hash, 64);
sph_echo512_close(&ctx.echo, hashB);
#endif
// 12 Hamsi
#if defined(__AES__)
update_and_final_groestl( &ctx.groestl, (char*)hash,
(const char*)hash, 512 );
#else
sph_groestl512 (&ctx.groestl, hash, 64);
sph_groestl512_close(&ctx.groestl, hash);
#endif
sph_skein512(&ctx.skein, hash, 64);
sph_skein512_close(&ctx.skein, hash);
sph_jh512(&ctx.jh, hash, 64);
sph_jh512_close(&ctx.jh, hash);
sph_keccak512(&ctx.keccak, hash, 64);
sph_keccak512_close(&ctx.keccak, hash);
sph_hamsi512(&ctx.hamsi, hashB, 64);
sph_hamsi512_close(&ctx.hamsi, hash);
asm volatile ("emms");
memcpy(output, hashB, 32);
}