v3.9.8

algo/x16/x16r-4way.c

@@ -68,21 +68,7 @@ void x16r_4way_hash( void* output, const void* input )
    int size = 80;
 
    dintrlv_4x64( hash0, hash1, hash2, hash3, input, 640 );
-/* 
-   if ( s_ntime == UINT32_MAX )
-   {
-      const uint8_t* tmp = (uint8_t*) in0;
-      x16_r_s_getAlgoString( &tmp[4], hashOrder );
-   }
-*/
-   // Input data is both 64 bit interleaved (input)
-   // and deinterleaved in inp0-3.
-   // If First function uses 64 bit data it is not required to interleave inp
-   // first. It may use the inerleaved data dmost convenient, ie 4way 64 bit.
-   // All other functions assume data is deinterleaved in hash0-3
-   // All functions must exit with data deinterleaved in hash0-3.
-   // Alias in0-3 points to either inp0-3 or hash0-3 according to
-   // its hashOrder position. Size is also set accordingly.
+
    for ( int i = 0; i < 16; i++ )
    {
       const char elem = hashOrder[i];

algo/x16/x16r-gate.c

@@ -42,8 +42,23 @@ bool register_x16r_algo( algo_gate_t* gate )
   gate->hash      = (void*)&x16r_hash;
 #endif
   gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT;
-  gate->set_target = (void*)&alt_set_target;
   x16_r_s_getAlgoString = (void*)&x16r_getAlgoString;
+  opt_target_factor = 256.0;
+  return true;
+};
+
+bool register_x16rv2_algo( algo_gate_t* gate )
+{
+#if defined (X16R_4WAY)
+  gate->scanhash  = (void*)&scanhash_x16rv2_4way;
+  gate->hash      = (void*)&x16rv2_4way_hash;
+#else
+  gate->scanhash  = (void*)&scanhash_x16rv2;
+  gate->hash      = (void*)&x16rv2_hash;
+#endif
+  gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT;
+  x16_r_s_getAlgoString = (void*)&x16r_getAlgoString;
+  opt_target_factor = 256.0;
   return true;
 };
 
@@ -57,8 +72,8 @@ bool register_x16s_algo( algo_gate_t* gate )
   gate->hash      = (void*)&x16r_hash;
 #endif
   gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT;
-  gate->set_target = (void*)&alt_set_target;
   x16_r_s_getAlgoString = (void*)&x16s_getAlgoString;
+  opt_target_factor = 256.0;
   return true;
 };
 
@@ -189,7 +204,7 @@ bool register_x16rt_algo( algo_gate_t* gate )
   gate->hash      = (void*)&x16rt_hash;
 #endif
   gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT;
-  gate->set_target = (void*)&alt_set_target;
+  opt_target_factor = 256.0;
   return true;
 };
 
@@ -203,8 +218,8 @@ bool register_x16rt_veil_algo( algo_gate_t* gate )
   gate->hash      = (void*)&x16rt_hash;
 #endif
   gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT;
-  gate->set_target = (void*)&alt_set_target;
   gate->build_extraheader = (void*)&veil_build_extraheader;
+  opt_target_factor = 256.0;
   return true;
 };
 
@@ -212,19 +227,13 @@ bool register_x16rt_veil_algo( algo_gate_t* gate )
 //
 //    HEX
 
-
-void hex_set_target( struct work* work, double job_diff )
-{
-   work_set_target( work, job_diff / (128.0 * opt_diff_factor) );
-}
-
 bool register_hex_algo( algo_gate_t* gate )
 {
   gate->scanhash        = (void*)&scanhash_hex;
   gate->hash            = (void*)&hex_hash;
   gate->optimizations   = SSE2_OPT | AES_OPT | AVX2_OPT;
   gate->gen_merkle_root = (void*)&SHA256_gen_merkle_root;
-  gate->set_target      = (void*)&hex_set_target;
+  opt_target_factor = 128.0;
   return true;
 };
 
@@ -244,8 +253,8 @@ bool register_x21s_algo( algo_gate_t* gate )
   gate->miner_thread_init = (void*)&x21s_thread_init;
 #endif
   gate->optimizations     = SSE2_OPT | AES_OPT | AVX2_OPT | SHA_OPT;
-  gate->set_target        = (void*)&alt_set_target;
   x16_r_s_getAlgoString   = (void*)&x16s_getAlgoString;
+  opt_target_factor = 256.0;
   return true;
 };
 

algo/x16/x16r-gate.h

@@ -38,6 +38,7 @@ void x16rt_getAlgoString( const uint32_t *timeHash, char *output );
 void x16rt_getTimeHash( const uint32_t timeStamp, void* timeHash );
 
 bool register_x16r_algo( algo_gate_t* gate );
+bool register_x16rv2_algo( algo_gate_t* gate );
 bool register_x16s_algo( algo_gate_t* gate );
 bool register_x16rt_algo( algo_gate_t* gate );
 bool register_hex__algo( algo_gate_t* gate );
@@ -49,6 +50,10 @@ void x16r_4way_hash( void *state, const void *input );
 int scanhash_x16r_4way( struct work *work, uint32_t max_nonce,
                         uint64_t *hashes_done, struct thr_info *mythr );
 
+void x16rv2_4way_hash( void *state, const void *input );
+int scanhash_x16rv2_4way( struct work *work, uint32_t max_nonce,
+                        uint64_t *hashes_done, struct thr_info *mythr );
+
 void x16rt_4way_hash( void *state, const void *input );
 int scanhash_x16rt_4way( struct work *work, uint32_t max_nonce,
                         uint64_t *hashes_done, struct thr_info *mythr );
@@ -64,6 +69,10 @@ void x16r_hash( void *state, const void *input );
 int scanhash_x16r( struct work *work, uint32_t max_nonce,
                    uint64_t *hashes_done, struct thr_info *mythr );
 
+void x16rv2_hash( void *state, const void *input );
+int scanhash_x16rv2( struct work *work, uint32_t max_nonce,
+                   uint64_t *hashes_done, struct thr_info *mythr );
+
 void x16rt_hash( void *state, const void *input );
 int scanhash_x16rt( struct work *work, uint32_t max_nonce,
                    uint64_t *hashes_done, struct thr_info *mythr );

algo/x16/x16rv2-4way.c

@@ -0,0 +1,384 @@
+/**
+ * x16r algo implementation
+ *
+ * Implementation by tpruvot@github Jan 2018
+ * Optimized by JayDDee@github Jan 2018
+ */
+#include "x16r-gate.h"
+
+#if defined (X16R_4WAY)
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include "algo/blake/blake-hash-4way.h"
+#include "algo/bmw/bmw-hash-4way.h"
+#include "algo/groestl/aes_ni/hash-groestl.h"
+#include "algo/groestl/aes_ni/hash-groestl.h"
+#include "algo/skein/skein-hash-4way.h"
+#include "algo/jh/jh-hash-4way.h"
+#include "algo/keccak/keccak-hash-4way.h"
+#include "algo/shavite/sph_shavite.h"
+#include "algo/luffa/luffa-hash-2way.h"
+#include "algo/cubehash/cubehash_sse2.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"
+#include "algo/shabal/shabal-hash-4way.h"
+#include "algo/whirlpool/sph_whirlpool.h"
+#include "algo/sha/sha-hash-4way.h"
+#include "algo/tiger/sph_tiger.h"
+
+static __thread uint32_t s_ntime = UINT32_MAX;
+static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
+
+union _x16rv2_4way_context_overlay
+{
+    blake512_4way_context   blake;
+    bmw512_4way_context     bmw;
+    hashState_echo          echo;
+    hashState_groestl       groestl;
+    skein512_4way_context   skein;
+    jh512_4way_context      jh;
+    keccak512_4way_context  keccak;
+    luffa_2way_context      luffa;
+    cubehashParam           cube;
+    sph_shavite512_context  shavite;
+    simd_2way_context       simd;
+    hamsi512_4way_context   hamsi;
+    sph_fugue512_context    fugue;
+    shabal512_4way_context  shabal;
+    sph_whirlpool_context   whirlpool;
+    sha512_4way_context     sha512;
+    sph_tiger_context       tiger;
+};
+typedef union _x16rv2_4way_context_overlay x16rv2_4way_context_overlay;
+
+// Pad the 24 bytes tiger hash to 64 bytes
+inline void padtiger512( uint32_t* hash )
+{
+  for ( int i = 6; i < 16; i++ ) hash[i] = 0;
+}
+
+void x16rv2_4way_hash( void* output, const void* input )
+{
+   uint32_t hash0[24] __attribute__ ((aligned (64)));
+   uint32_t hash1[24] __attribute__ ((aligned (64)));
+   uint32_t hash2[24] __attribute__ ((aligned (64)));
+   uint32_t hash3[24] __attribute__ ((aligned (64)));
+   uint32_t vhash[24*4] __attribute__ ((aligned (64)));
+   x16rv2_4way_context_overlay ctx;
+   void *in0 = (void*) hash0;
+   void *in1 = (void*) hash1;
+   void *in2 = (void*) hash2;
+   void *in3 = (void*) hash3;
+   int size = 80;
+
+   dintrlv_4x64( hash0, hash1, hash2, hash3, input, 640 );
+
+   for ( int i = 0; i < 16; i++ )
+   {
+      const char elem = hashOrder[i];
+      const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
+
+      switch ( algo )
+      {
+         case BLAKE:
+            blake512_4way_init( &ctx.blake );
+            if ( i == 0 )
+               blake512_4way( &ctx.blake, input, size );
+            else
+            {
+               intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
+               blake512_4way( &ctx.blake, vhash, size );
+            }
+            blake512_4way_close( &ctx.blake, vhash );
+            dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
+         break;
+         case BMW:
+            bmw512_4way_init( &ctx.bmw );
+            if ( i == 0 )
+               bmw512_4way( &ctx.bmw, input, size );
+            else
+            {
+               intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
+               bmw512_4way( &ctx.bmw, vhash, size );
+            }
+            bmw512_4way_close( &ctx.bmw, vhash );
+            dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
+         break;
+         case GROESTL:
+               init_groestl( &ctx.groestl, 64 );
+               update_and_final_groestl( &ctx.groestl, (char*)hash0,
+                                                 (const char*)in0, size<<3 );
+               init_groestl( &ctx.groestl, 64 );
+               update_and_final_groestl( &ctx.groestl, (char*)hash1,
+                                                 (const char*)in1, size<<3 );
+               init_groestl( &ctx.groestl, 64 );
+               update_and_final_groestl( &ctx.groestl, (char*)hash2,
+                                                 (const char*)in2, size<<3 );
+               init_groestl( &ctx.groestl, 64 );
+               update_and_final_groestl( &ctx.groestl, (char*)hash3,
+                                                 (const char*)in3, size<<3 );
+         break;
+         case SKEIN:
+            skein512_4way_init( &ctx.skein );
+            if ( i == 0 )
+               skein512_4way( &ctx.skein, input, size );
+            else
+            {
+               intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
+               skein512_4way( &ctx.skein, vhash, size );
+            }
+            skein512_4way_close( &ctx.skein, vhash );
+            dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
+         break;
+         case JH:
+            jh512_4way_init( &ctx.jh );
+            if ( i == 0 )
+               jh512_4way( &ctx.jh, input, size );
+            else
+            {
+               intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
+               jh512_4way( &ctx.jh, vhash, size );
+            }
+            jh512_4way_close( &ctx.jh, vhash );
+            dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
+         break;
+         case KECCAK:
+             sph_tiger_init( &ctx.tiger );
+			    sph_tiger( &ctx.tiger, in0, size );
+			    sph_tiger_close( &ctx.tiger, hash0 );
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in1, size );
+             sph_tiger_close( &ctx.tiger, hash1 );
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in2, size );
+             sph_tiger_close( &ctx.tiger, hash2 );
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in3, size );
+             sph_tiger_close( &ctx.tiger, hash3 );
+
+             for ( int i = (24/4); i < (64/4); i++ )
+                hash0[i] = hash1[i] = hash2[i] = hash3[i] = 0;
+
+             intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
+             keccak512_4way_init( &ctx.keccak );
+             keccak512_4way( &ctx.keccak, vhash, 64 );
+             keccak512_4way_close( &ctx.keccak, vhash );
+             dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
+         break;
+         case LUFFA:
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in0, size );
+             sph_tiger_close( &ctx.tiger, hash0 );
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in1, size );
+             sph_tiger_close( &ctx.tiger, hash1 );
+
+             for ( int i = (24/4); i < (64/4); i++ )
+                hash0[i] = hash1[i] = 0;
+
+             intrlv_2x128( vhash, hash0, hash1, 512 );
+             luffa_2way_init( &ctx.luffa, 512 );
+             luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
+             dintrlv_2x128( hash0, hash1, vhash, 512 );
+
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in2, size );
+             sph_tiger_close( &ctx.tiger, hash2 );
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in3, size );
+             sph_tiger_close( &ctx.tiger, hash3 );
+
+             for ( int i = (24/4); i < (64/4); i++ )
+                hash2[i] = hash3[i] = 0;
+             
+             intrlv_2x128( vhash, hash2, hash3, 512 );
+             luffa_2way_init( &ctx.luffa, 512 );
+             luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
+             dintrlv_2x128( hash2, hash3, vhash, 512 );
+         break;
+         case CUBEHASH:
+             cubehashInit( &ctx.cube, 512, 16, 32 );
+             cubehashUpdateDigest( &ctx.cube, (byte*) hash0,
+                                   (const byte*)in0, size );
+             cubehashInit( &ctx.cube, 512, 16, 32 );
+             cubehashUpdateDigest( &ctx.cube, (byte*) hash1,
+                                   (const byte*)in1, size );
+             cubehashInit( &ctx.cube, 512, 16, 32 );
+             cubehashUpdateDigest( &ctx.cube, (byte*) hash2,
+                                   (const byte*)in2, size );
+             cubehashInit( &ctx.cube, 512, 16, 32 );
+             cubehashUpdateDigest( &ctx.cube, (byte*) hash3,
+                                   (const byte*)in3, size );
+         break;
+         case SHAVITE:
+             sph_shavite512_init( &ctx.shavite );
+             sph_shavite512( &ctx.shavite, in0, size );
+             sph_shavite512_close( &ctx.shavite, hash0 );
+             sph_shavite512_init( &ctx.shavite );
+             sph_shavite512( &ctx.shavite, in1, size );
+             sph_shavite512_close( &ctx.shavite, hash1 );
+             sph_shavite512_init( &ctx.shavite );
+             sph_shavite512( &ctx.shavite, in2, size );
+             sph_shavite512_close( &ctx.shavite, hash2 );
+             sph_shavite512_init( &ctx.shavite );
+             sph_shavite512( &ctx.shavite, in3, size );
+             sph_shavite512_close( &ctx.shavite, hash3 );
+         break;
+         case SIMD:
+             intrlv_2x128( vhash, in0, in1, size<<3 );
+             simd_2way_init( &ctx.simd, 512 );
+             simd_2way_update_close( &ctx.simd, vhash, vhash, size<<3 );
+             dintrlv_2x128( hash0, hash1, vhash, 512 );
+             intrlv_2x128( vhash, in2, in3, size<<3 );
+             simd_2way_init( &ctx.simd, 512 );
+             simd_2way_update_close( &ctx.simd, vhash, vhash, size<<3 );
+             dintrlv_2x128( hash2, hash3, vhash, 512 );
+         break;
+         case ECHO:
+             init_echo( &ctx.echo, 512 );
+             update_final_echo ( &ctx.echo, (BitSequence *)hash0,
+                                (const BitSequence*)in0, size<<3 );
+             init_echo( &ctx.echo, 512 );
+             update_final_echo ( &ctx.echo, (BitSequence *)hash1,
+                                (const BitSequence*)in1, size<<3 );
+             init_echo( &ctx.echo, 512 );
+             update_final_echo ( &ctx.echo, (BitSequence *)hash2,
+                                (const BitSequence*)in2, size<<3 );
+             init_echo( &ctx.echo, 512 );
+             update_final_echo ( &ctx.echo, (BitSequence *)hash3,
+                                (const BitSequence*)in3, size<<3 );
+         break;
+         case HAMSI:
+             intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
+             hamsi512_4way_init( &ctx.hamsi );
+             hamsi512_4way( &ctx.hamsi, vhash, size );
+             hamsi512_4way_close( &ctx.hamsi, vhash );
+             dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
+         break;
+         case FUGUE:
+             sph_fugue512_init( &ctx.fugue );
+             sph_fugue512( &ctx.fugue, in0, size );
+             sph_fugue512_close( &ctx.fugue, hash0 );
+             sph_fugue512_init( &ctx.fugue );
+             sph_fugue512( &ctx.fugue, in1, size );
+             sph_fugue512_close( &ctx.fugue, hash1 );
+             sph_fugue512_init( &ctx.fugue );
+             sph_fugue512( &ctx.fugue, in2, size );
+             sph_fugue512_close( &ctx.fugue, hash2 );
+             sph_fugue512_init( &ctx.fugue );
+             sph_fugue512( &ctx.fugue, in3, size );
+             sph_fugue512_close( &ctx.fugue, hash3 );
+         break;
+         case SHABAL:
+             intrlv_4x32( vhash, in0, in1, in2, in3, size<<3 );
+             shabal512_4way_init( &ctx.shabal );
+             shabal512_4way( &ctx.shabal, vhash, size );
+             shabal512_4way_close( &ctx.shabal, vhash );
+             dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, 512 );
+         break;
+         case WHIRLPOOL:
+             sph_whirlpool_init( &ctx.whirlpool );
+             sph_whirlpool( &ctx.whirlpool, in0, size );
+             sph_whirlpool_close( &ctx.whirlpool, hash0 );
+             sph_whirlpool_init( &ctx.whirlpool );
+             sph_whirlpool( &ctx.whirlpool, in1, size );
+             sph_whirlpool_close( &ctx.whirlpool, hash1 );
+             sph_whirlpool_init( &ctx.whirlpool );
+             sph_whirlpool( &ctx.whirlpool, in2, size );
+             sph_whirlpool_close( &ctx.whirlpool, hash2 );
+             sph_whirlpool_init( &ctx.whirlpool );
+             sph_whirlpool( &ctx.whirlpool, in3, size );
+             sph_whirlpool_close( &ctx.whirlpool, hash3 );
+         break;
+         case SHA_512:
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in0, size );
+             sph_tiger_close( &ctx.tiger, hash0 );
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in1, size );
+             sph_tiger_close( &ctx.tiger, hash1 );
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in2, size );
+             sph_tiger_close( &ctx.tiger, hash2 );
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in3, size );
+             sph_tiger_close( &ctx.tiger, hash3 );
+
+             for ( int i = (24/4); i < (64/4); i++ )
+                hash0[i] = hash1[i] = hash2[i] = hash3[i] = 0;
+ 
+             intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
+             sha512_4way_init( &ctx.sha512 );
+             sha512_4way( &ctx.sha512, vhash, 64 );
+             sha512_4way_close( &ctx.sha512, vhash );
+             dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
+         break;
+      }
+      size = 64;
+   }
+   memcpy( output,    hash0, 32 );
+   memcpy( output+32, hash1, 32 );
+   memcpy( output+64, hash2, 32 );
+   memcpy( output+96, hash3, 32 );
+}
+
+int scanhash_x16rv2_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 endiandata[20] __attribute__((aligned(64)));
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   const uint32_t Htarg = ptarget[7];
+   const uint32_t first_nonce = pdata[19];
+   uint32_t n = first_nonce;
+   int thr_id = mythr->id;  // thr_id arg is deprecated
+    __m256i  *noncev = (__m256i*)vdata + 9;   // aligned
+   volatile uint8_t *restart = &(work_restart[thr_id].restart);
+
+   casti_m256i( endiandata, 0 ) = mm256_bswap_32( casti_m256i( pdata, 0 ) );
+   casti_m256i( endiandata, 1 ) = mm256_bswap_32( casti_m256i( pdata, 1 ) );
+   casti_m128i( endiandata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
+
+   if ( s_ntime != endiandata[17] )
+   {
+      uint32_t ntime = swab32(pdata[17]);
+      x16_r_s_getAlgoString( (const uint8_t*) (&endiandata[1]), hashOrder );
+      s_ntime = ntime;
+      if ( opt_debug && !thr_id )
+              applog( LOG_DEBUG, "hash order %s (%08x)", hashOrder, ntime );
+   }
+
+   if ( opt_benchmark )
+      ptarget[7] = 0x0cff;
+
+   uint64_t *edata = (uint64_t*)endiandata;
+   intrlv_4x64( (uint64_t*)vdata, edata, edata, edata, edata, 640 );
+
+   do
+   {
+      *noncev = mm256_intrlv_blend_32( mm256_bswap_32(
+               _mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ) ), *noncev );
+
+      x16rv2_4way_hash( hash, vdata );
+      pdata[19] = n;
+
+      for ( int i = 0; i < 4; i++ )  if ( (hash+(i<<3))[7] <= Htarg )
+      if( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
+      {
+         pdata[19] = n+i;
+         submit_lane_solution( work, hash+(i<<3), mythr, i );
+      }
+      n += 4;
+   } while ( (  n < max_nonce ) && !(*restart) );
+
+   *hashes_done = n - first_nonce + 1;
+   return 0;
+}
+
+#endif

algo/x16/x16rv2.c

@@ -0,0 +1,247 @@
+/**
+ * x16r algo implementation
+ *
+ * Implementation by tpruvot@github Jan 2018
+ * Optimized by JayDDee@github Jan 2018
+ */
+#include "x16r-gate.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include "algo/blake/sph_blake.h"
+#include "algo/bmw/sph_bmw.h"
+#include "algo/groestl/sph_groestl.h"
+#include "algo/jh/sph_jh.h"
+#include "algo/keccak/sph_keccak.h"
+#include "algo/skein/sph_skein.h"
+#include "algo/shavite/sph_shavite.h"
+#include "algo/luffa/luffa_for_sse2.h"
+#include "algo/cubehash/cubehash_sse2.h"
+#include "algo/simd/nist.h"
+#include "algo/echo/sph_echo.h"
+#include "algo/hamsi/sph_hamsi.h"
+#include "algo/fugue/sph_fugue.h"
+#include "algo/shabal/sph_shabal.h"
+#include "algo/whirlpool/sph_whirlpool.h"
+#include <openssl/sha.h>
+#include "algo/tiger/sph_tiger.h"
+#if defined(__AES__)
+  #include "algo/echo/aes_ni/hash_api.h"
+  #include "algo/groestl/aes_ni/hash-groestl.h"
+#endif
+
+static __thread uint32_t s_ntime = UINT32_MAX;
+static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
+
+union _x16rv2_context_overlay
+{
+#if defined(__AES__)
+        hashState_echo          echo;
+        hashState_groestl       groestl;
+#else
+        sph_groestl512_context   groestl;
+        sph_echo512_context      echo;
+#endif
+        sph_blake512_context    blake;
+        sph_bmw512_context      bmw;
+        sph_skein512_context    skein;
+        sph_jh512_context       jh;
+        sph_keccak512_context   keccak;
+        hashState_luffa         luffa;
+        cubehashParam           cube;
+        sph_shavite512_context  shavite;
+        hashState_sd            simd;
+        sph_hamsi512_context    hamsi;
+        sph_fugue512_context    fugue;
+        sph_shabal512_context   shabal;
+        sph_whirlpool_context   whirlpool;
+        SHA512_CTX              sha512;
+        sph_tiger_context       tiger;
+};
+typedef union _x16rv2_context_overlay x16rv2_context_overlay;
+
+// Pad the 24 bytes tiger hash to 64 bytes
+inline void padtiger512(uint32_t* hash) {
+   for (int i = (24/4); i < (64/4); i++) hash[i] = 0;
+}
+
+void x16rv2_hash( void* output, const void* input )
+{
+   uint32_t _ALIGN(128) hash[16];
+   x16rv2_context_overlay ctx;
+   void *in = (void*) input;
+   int size = 80;
+/*
+   if ( s_ntime == UINT32_MAX )
+   {
+      const uint8_t* in8 = (uint8_t*) input;
+      x16_r_s_getAlgoString( &in8[4], hashOrder );
+   }
+*/
+   for ( int i = 0; i < 16; i++ )
+   {
+      const char elem = hashOrder[i];
+      const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
+
+      switch ( algo )
+      {
+         case BLAKE:
+            sph_blake512_init( &ctx.blake );
+            sph_blake512( &ctx.blake, in, size );
+            sph_blake512_close( &ctx.blake, hash );
+         break;
+         case BMW:
+            sph_bmw512_init( &ctx.bmw );
+            sph_bmw512(&ctx.bmw, in, size);
+            sph_bmw512_close(&ctx.bmw, hash);
+         break;
+         case GROESTL:
+#if defined(__AES__)
+            init_groestl( &ctx.groestl, 64 );
+            update_and_final_groestl( &ctx.groestl, (char*)hash,
+                                      (const char*)in, size<<3 );
+#else
+            sph_groestl512_init( &ctx.groestl );
+            sph_groestl512( &ctx.groestl, in, size );
+            sph_groestl512_close(&ctx.groestl, hash);
+#endif
+         break;
+         case SKEIN:
+            sph_skein512_init( &ctx.skein );
+            sph_skein512( &ctx.skein, in, size );
+            sph_skein512_close( &ctx.skein, hash );
+         break;
+         case JH:
+            sph_jh512_init( &ctx.jh );
+            sph_jh512(&ctx.jh, in, size );
+            sph_jh512_close(&ctx.jh, hash );
+         break;
+         case KECCAK:
+            sph_tiger_init( &ctx.tiger );
+            sph_tiger( &ctx.tiger, in, size );
+            sph_tiger_close( &ctx.tiger, hash );
+            padtiger512( hash );
+            sph_keccak512_init( &ctx.keccak );
+            sph_keccak512( &ctx.keccak, hash, 64 );
+            sph_keccak512_close( &ctx.keccak, hash );
+         break;
+         case LUFFA:
+            sph_tiger_init( &ctx.tiger );
+            sph_tiger( &ctx.tiger, in, size );
+            sph_tiger_close( &ctx.tiger, hash );
+            padtiger512( hash );
+            init_luffa( &ctx.luffa, 512 );
+            update_and_final_luffa( &ctx.luffa, (BitSequence*)hash,
+                                    (const BitSequence*)hash, 64 );
+         break;
+         case CUBEHASH:
+            cubehashInit( &ctx.cube, 512, 16, 32 );
+            cubehashUpdateDigest( &ctx.cube, (byte*) hash,
+                                  (const byte*)in, size );
+         break;
+         case SHAVITE:
+            sph_shavite512_init( &ctx.shavite );
+            sph_shavite512( &ctx.shavite, in, size );
+            sph_shavite512_close( &ctx.shavite, hash );
+         break;
+         case SIMD:
+             init_sd( &ctx.simd, 512 );
+             update_final_sd( &ctx.simd, (BitSequence *)hash,
+                              (const BitSequence*)in, size<<3 );
+         break;
+         case ECHO:
+#if defined(__AES__)
+             init_echo( &ctx.echo, 512 );
+             update_final_echo ( &ctx.echo, (BitSequence *)hash,
+                                (const BitSequence*)in, size<<3 );
+#else
+             sph_echo512_init( &ctx.echo );
+             sph_echo512( &ctx.echo, in, size );
+             sph_echo512_close( &ctx.echo, hash );
+#endif
+         break;
+         case HAMSI:
+             sph_hamsi512_init( &ctx.hamsi );
+             sph_hamsi512( &ctx.hamsi, in, size );
+             sph_hamsi512_close( &ctx.hamsi, hash );
+         break;
+         case FUGUE:
+             sph_fugue512_init( &ctx.fugue );
+             sph_fugue512( &ctx.fugue, in, size );
+             sph_fugue512_close( &ctx.fugue, hash );
+         break;
+         case SHABAL:
+             sph_shabal512_init( &ctx.shabal );
+             sph_shabal512( &ctx.shabal, in, size );
+             sph_shabal512_close( &ctx.shabal, hash );
+         break;
+         case WHIRLPOOL:
+             sph_whirlpool_init( &ctx.whirlpool );
+             sph_whirlpool( &ctx.whirlpool, in, size );
+             sph_whirlpool_close( &ctx.whirlpool, hash );
+         break;
+         case SHA_512:
+             sph_tiger_init( &ctx.tiger );
+             sph_tiger( &ctx.tiger, in, size );
+             sph_tiger_close( &ctx.tiger, hash );
+             padtiger512( hash );
+             SHA512_Init( &ctx.sha512 );
+             SHA512_Update( &ctx.sha512, hash, 64 );
+             SHA512_Final( (unsigned char*) hash, &ctx.sha512 );
+         break;
+      }
+      in = (void*) hash;
+      size = 64;
+   }
+   memcpy(output, hash, 32);
+}
+
+int scanhash_x16rv2( struct work *work, uint32_t max_nonce,
+                   uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t _ALIGN(128) hash32[8];
+   uint32_t _ALIGN(128) endiandata[20];
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   const uint32_t Htarg = ptarget[7];
+   const uint32_t first_nonce = pdata[19];
+   int thr_id = mythr->id;  // thr_id arg is deprecated
+   uint32_t nonce = first_nonce;
+   volatile uint8_t *restart = &(work_restart[thr_id].restart);
+
+   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 ) );
+
+   if ( s_ntime != pdata[17] )
+   {
+      uint32_t ntime = swab32(pdata[17]);
+      x16_r_s_getAlgoString( (const uint8_t*) (&endiandata[1]), hashOrder );
+      s_ntime = ntime;
+      if ( opt_debug && !thr_id )
+              applog( LOG_DEBUG, "hash order %s (%08x)", hashOrder, ntime );
+   }
+
+   if ( opt_benchmark )
+      ptarget[7] = 0x0cff;
+
+   do
+   {
+      be32enc( &endiandata[19], nonce );
+      x16rv2_hash( hash32, endiandata );
+
+      if ( hash32[7] <= Htarg )
+      if (fulltest( hash32, ptarget ) && !opt_benchmark )
+      {
+         pdata[19] = nonce;
+         submit_solution( work, hash32, mythr );
+      }
+      nonce++;
+   } while ( nonce < max_nonce && !(*restart) );
+   pdata[19] = nonce;
+   *hashes_done = pdata[19] - first_nonce + 1;
+   return 0;
+}