v3.9.11

Makefile.am

@@ -18,7 +18,6 @@ dist_man_MANS	= cpuminer.1
 cpuminer_SOURCES = \
   cpu-miner.c \
   util.c \
-  uint256.cpp \
   api.c \
   sysinfos.c \
   algo-gate-api.c\
@@ -51,12 +50,15 @@ cpuminer_SOURCES = \
   algo/blake/blake.c \
   algo/blake/blake-4way.c \
   algo/blake/sph_blake2b.c \
-  algo/blake/blake2b.c \
   algo/blake/sph-blake2s.c \
   algo/blake/blake2s-hash-4way.c \
   algo/blake/blake2s.c \
   algo/blake/blake2s-gate.c \
   algo/blake/blake2s-4way.c \
+  algo/blake/blake2b-hash-4way.c \
+  algo/blake/blake2b.c \
+  algo/blake/blake2b-gate.c \
+  algo/blake/blake2b-4way.c \
   algo/blake/blakecoin-gate.c \
   algo/blake/mod_blakecoin.c \
   algo/blake/blakecoin.c \
@@ -115,6 +117,7 @@ cpuminer_SOURCES = \
   algo/keccak/keccak-4way.c\
   algo/keccak/keccak-gate.c \
   algo/keccak/sse2/keccak.c \
+  algo/lanehash/lane.c \
   algo/luffa/sph_luffa.c \
   algo/luffa/luffa.c \
   algo/luffa/luffa_for_sse2.c \
@@ -169,7 +172,8 @@ cpuminer_SOURCES = \
   algo/scryptjane/scrypt-jane.c \
   algo/sha/sph_sha2.c \
   algo/sha/sph_sha2big.c \
-  algo/sha/sha2-hash-4way.c \
+  algo/sha/sha256-hash-4way.c \
+  algo/sha/sha512-hash-4way.c \
   algo/sha/sha256_hash_11way.c \
   algo/sha/sha2.c \
   algo/sha/sha256t-gate.c \
@@ -197,6 +201,7 @@ cpuminer_SOURCES = \
   algo/skein/skein2-gate.c \
   algo/sm3/sm3.c \
   algo/sm3/sm3-hash-4way.c \
+  algo/swifftx/swifftx.c \
   algo/tiger/sph_tiger.c \
   algo/whirlpool/sph_whirlpool.c \
   algo/whirlpool/whirlpool-hash-4way.c \
@@ -259,8 +264,13 @@ cpuminer_SOURCES = \
   algo/x16/x16r-gate.c \
   algo/x16/x16r.c \
   algo/x16/x16r-4way.c \
+  algo/x16/x16rv2.c \
+  algo/x16/x16rv2-4way.c \
   algo/x16/x16rt.c \
   algo/x16/x16rt-4way.c \
+  algo/x16/hex.c \
+  algo/x16/x21s-4way.c \
+  algo/x16/x21s.c \
   algo/x17/x17-gate.c \
   algo/x17/x17.c \
   algo/x17/x17-4way.c \
@@ -271,10 +281,17 @@ cpuminer_SOURCES = \
   algo/x17/sonoa-4way.c \
   algo/x17/sonoa.c \
   algo/x20/x20r.c \
+  algo/x22/x22i-4way.c \
+  algo/x22/x22i.c \
+  algo/x22/x22i-gate.c \
+  algo/x22/x25x.c \
+  algo/x22/x25x-4way.c \
   algo/yescrypt/yescrypt.c \
   algo/yescrypt/sha256_Y.c \
   algo/yescrypt/yescrypt-best.c \
-  algo/yespower/yespower.c \
+  algo/yespower/yespower-gate.c \
+  algo/yespower/yespower-blake2b.c \
+  algo/yespower/crypto/blake2b-yp.c \
   algo/yespower/sha256_p.c \
   algo/yespower/yespower-opt.c
 

README.md

@@ -24,7 +24,7 @@ Requirements
 
 1. A x86_64 architecture CPU with a minimum of SSE2 support. This includes
 Intel Core2 and newer and AMD equivalents. In order to take advantage of AES_NI
-optimizations a CPU with AES_NI is required. This includes Intel Westbridge
+optimizations a CPU with AES_NI is required. This includes Intel Westmere
 and newer and AMD equivalents. Further optimizations are available on some
 algoritms for CPUs with AVX and AVX2, Sandybridge and Haswell respectively.
 
@@ -55,8 +55,9 @@ Supported Algorithms
                           axiom         Shabal-256 MemoHash
                           bastion
                           blake         Blake-256 (SFR)
-                          blakecoin     blake256r8
+                          blake2b       Blake2b 256
                           blake2s       Blake-2 S
+                          blakecoin     blake256r8
                           bmw           BMW 256
                           bmw512        BMW 512
                           c11           Chaincoin
@@ -67,6 +68,7 @@ Supported Algorithms
                           fresh         Fresh
                           groestl       Groestl coin
                           heavy         Heavy
+                          hex           x16r-hex
                           hmq1725       Espers
                           hodl          Hodlcoin
                           jha           Jackpotcoin
@@ -85,10 +87,12 @@ Supported Algorithms
                           neoscrypt     NeoScrypt(128, 2, 1)
                           nist5         Nist5
                           pentablake    Pentablake
-                          phi1612       phi, LUX coin (original algo)
-                          phi2          LUX coin (new algo)
+                          phi1612       phi
+                          phi2          Luxcoin (LUX)
+                          phi2-lux      identical to phi2
                           pluck         Pluck:128 (Supcoin)
                           polytimos     Ninja
+                          power2b       MicroBitcoin (MBC)
                           quark         Quark
                           qubit         Qubit
                           scrypt        scrypt(1024, 1, 1) (default)
@@ -118,11 +122,15 @@ Supported Algorithms
                           x13sm3        hsr (Hshare)
                           x14           X14
                           x15           X15
-                          x16r          Ravencoin (RVN)
+                          x16r          
+                          x16rv2        Ravencoin (RVN)
                           x16rt         Gincoin (GIN)
-                          x16rt_veil    Veil (VEIL)
+                          x16rt-veil    Veil (VEIL)
                           x16s          Pigeoncoin (PGN)
                           x17
+                          x21s
+                          x22i
+                          x25x
                           xevan         Bitsend (BSD)
                           yescrypt      Globalboost-Y (BSTY)
                           yescryptr8    BitZeny (ZNY)
@@ -130,6 +138,7 @@ Supported Algorithms
                           yescryptr32   WAVI
                           yespower      Cryply
                           yespowerr16   Yenten (YTN)
+                          yespower-b2b  generic yespower + blake2b
                           zr5           Ziftr
 
 Errata
@@ -152,14 +161,17 @@ Benchmark testing does not work for x11evo.
 Bugs
 ----
 
-Users are encouraged to post their bug reports on the Bitcoin Talk
-forum at:
+Users are encouraged to post their bug reports using git issues or on the
+Bitcoin Talk forum or opening an issue in git:
 
 https://bitcointalk.org/index.php?topic=1326803.0
 
-All problem reports must be accompanied by a proper definition.
+https://github.com/JayDDee/cpuminer-opt/issues
+
+All problem reports must be accompanied by a proper problem definition.
 This should include how the problem occurred, the command line and
-output from the miner showing the startup and any errors.
+output from the miner showing the startup messages and any errors.
+A history is also useful, ie did it work before.
 
 Donations
 ---------
@@ -167,10 +179,6 @@ Donations
 cpuminer-opt has no fees of any kind but donations are accepted.
 
  BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT
- ETH: 0x72122edabcae9d3f57eab0729305a425f6fef6d0
- LTC: LdUwoHJnux9r9EKqFWNvAi45kQompHk6e8
- BCH: 1QKYkB6atn4P7RFozyziAXLEnurwnUM1cQ
- BTG: GVUyECtRHeC5D58z9F3nGGfVQndwnsPnHQ
 
 Happy mining!
 

RELEASE_NOTES

@@ -1,14 +1,6 @@
 cpuminer-opt is a console program run from the command line using the
 keyboard, not the mouse.
 
-cpuminer-opt now supports HW SHA acceleration available on AMD Ryzen CPUs.
-This feature requires recent SW including GCC version 5 or higher and
-openssl version 1.1 or higher. It may also require using "-march=znver1"
-compile flag.
-
-cpuminer-opt is a console program, if you're using a mouse you're doing it
-wrong.
-
 Security warning
 ----------------
 
@@ -34,10 +26,94 @@ Intel Core2 or newer, or AMD Steamroller or newer CPU. ARM CPUs are not
 supported.
 
 64 bit Linux or Windows operating system. Apple and Android are not supported.
+FreeBSD YMMV.
 
 Change Log
 ----------
 
+v3.9.11
+
+Added x22i & x25x algos.
+Blake2s 2% faster AVX2 with Intel CPU, slower with Ryzen v1, v2 ?
+
+v3.9.10
+
+Faster X* algos with AVX2.
+Small improvements to summary stats report.
+
+v3.9.9.1
+
+Fixed a day1 bug that could cause the miner to idle for up to 2 minutes
+under certain circumstances.
+
+Redesigned summary stats report now includes session statistics.
+
+More robust handling of statistics to reduce corruption.
+
+Removed --hide-diff option.
+
+Better handling of cpu-affinity with more than 64 CPUs.
+
+v3.9.9
+
+Added power2b algo for MicroBitcoin.
+Added generic yespower-b2b (yespower + blake2b) algo to be used with
+the parameters introduced in v3.9.7 for yespower & yescrypt.
+Display additional info when a share is rejected.
+Some low level enhancements and minor tweaking of log output.
+RELEASE_NOTES (this file) and README.md  added to Windows release package.
+
+v3.9.8.1
+
+Summary log report will be generated on stratum diff change or after 5 minutes,
+whichever comes first, to prevent incorrect data in the report.
+
+Removed phi2-lux alias (introduced in v3.9.8) due to Luxcoin's planned fork
+to a new algo. The new Luxcoin algo is not supported by cpuminer-opt.
+Until the fork Luxcoin can be mined using phi2 algo.
+
+--hide-diff option is deprecated and has no effect. It will be removed in a
+future release.
+
+v3.9.8
+
+Changes to log output to provide data more relevant to actual mining
+performance.
+phi2 can now handle pools with a mix of coins that use and don't use roots.
+phi2-lux added as an alias for phi2 as they are identical except for roots.
+Add x16rv2 algo for Ravencoin fork.
+
+v3.9.7
+
+Command line option changes:
+
+"-R" is no longer used as a shortcut for "--retry-pause", users must
+use the long option.
+
+New options:
+
+-N, --param-n: set the N parameter for yescrypt, yespower or scrypt algos
+-R, --param-r: set the R parameter for yescrypt or yespower algos, scrypt is
+     hardcoded with R=1
+-K, --param-key: set the client key/pers parameter for yescrypt/yespower algos.
+
+These options can be used to mine yescrypt or yespower variations using
+the generic yescrypt or yespower algo name and specifying the parameters
+manually. They can even be used to mine variations that aren't formally
+supported by a unique algo name. Existing algos can continue to to be mined
+using their original name without parameters.
+
+v3.9.6.2
+
+New algo blake2b.
+Faster myr-gr on Ryzen using SHA.
+Faster blake2s SSE2.
+Small speedup of around 1% for several other algos.
+
+v3.9.6.1
+
+New algos: x21s, hex (alias x16r-hex).
+
 v3.9.6
 
 New algos: bmw512, x16rt, x16rt-veil (alias veil), x13bcd (alias bcd).

algo-gate-api.c

@@ -116,13 +116,10 @@ void init_algo_gate( algo_gate_t* gate )
    gate->get_nonceptr            = (void*)&std_get_nonceptr;
    gate->work_decode             = (void*)&std_le_work_decode;
    gate->decode_extra_data       = (void*)&do_nothing;
-   gate->wait_for_diff           = (void*)&std_wait_for_diff;
-   gate->get_max64               = (void*)&get_max64_0x1fffffLL;
    gate->gen_merkle_root         = (void*)&sha256d_gen_merkle_root;
    gate->stratum_gen_work        = (void*)&std_stratum_gen_work;
    gate->build_stratum_request   = (void*)&std_le_build_stratum_request;
    gate->malloc_txs_request      = (void*)&std_malloc_txs_request;
-   gate->set_target              = (void*)&std_set_target;
    gate->submit_getwork_result   = (void*)&std_le_submit_getwork_result;
    gate->build_block_header      = (void*)&std_build_block_header;
    gate->build_extraheader       = (void*)&std_build_extraheader;
@@ -167,9 +164,9 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
     case ALGO_AXIOM:         register_axiom_algo         ( gate ); break;
     case ALGO_BASTION:       register_bastion_algo       ( gate ); break;
     case ALGO_BLAKE:         register_blake_algo         ( gate ); break;
-    case ALGO_BLAKECOIN:     register_blakecoin_algo     ( gate ); break;
-//    case ALGO_BLAKE2B:      register_blake2b_algo     ( gate ); break;
+    case ALGO_BLAKE2B:       register_blake2b_algo       ( gate ); break;
     case ALGO_BLAKE2S:       register_blake2s_algo       ( gate ); break;
+    case ALGO_BLAKECOIN:     register_blakecoin_algo     ( gate ); break;
     case ALGO_BMW512:        register_bmw512_algo        ( gate ); break;
     case ALGO_C11:           register_c11_algo           ( gate ); break;
     case ALGO_CRYPTOLIGHT:   register_cryptolight_algo   ( gate ); break;
@@ -182,6 +179,7 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
     case ALGO_FRESH:         register_fresh_algo         ( gate ); break;
     case ALGO_GROESTL:       register_groestl_algo       ( gate ); break;
     case ALGO_HEAVY:         register_heavy_algo         ( gate ); break;
+    case ALGO_HEX:           register_hex_algo           ( gate ); break;
     case ALGO_HMQ1725:       register_hmq1725_algo       ( gate ); break;
     case ALGO_HODL:          register_hodl_algo          ( gate ); break;
     case ALGO_JHA:           register_jha_algo           ( gate ); break;
@@ -204,6 +202,7 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
     case ALGO_PHI2:          register_phi2_algo          ( gate ); break;
     case ALGO_PLUCK:         register_pluck_algo         ( gate ); break;
     case ALGO_POLYTIMOS:     register_polytimos_algo     ( gate ); break;
+    case ALGO_POWER2B:       register_power2b_algo       ( gate ); break;
     case ALGO_QUARK:         register_quark_algo         ( gate ); break;
     case ALGO_QUBIT:         register_qubit_algo         ( gate ); break;
     case ALGO_SCRYPT:        register_scrypt_algo        ( gate ); break;
@@ -233,10 +232,14 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
     case ALGO_X14:           register_x14_algo           ( gate ); break;
     case ALGO_X15:           register_x15_algo           ( gate ); break;
     case ALGO_X16R:          register_x16r_algo          ( gate ); break;
+    case ALGO_X16RV2:        register_x16rv2_algo        ( gate ); break;
     case ALGO_X16RT:         register_x16rt_algo         ( gate ); break;
     case ALGO_X16RT_VEIL:    register_x16rt_veil_algo    ( gate ); break;
     case ALGO_X16S:          register_x16s_algo          ( gate ); break;
     case ALGO_X17:           register_x17_algo           ( gate ); break;
+    case ALGO_X21S:          register_x21s_algo          ( gate ); break;
+    case ALGO_X22I:          register_x22i_algo          ( gate ); break;
+    case ALGO_X25X:          register_x25x_algo          ( gate ); break;
     case ALGO_XEVAN:         register_xevan_algo         ( gate ); break;
 /*    case ALGO_YESCRYPT:     register_yescrypt_05_algo     ( gate ); break;
      case ALGO_YESCRYPTR8:   register_yescryptr8_05_algo   ( gate ); break;
@@ -249,6 +252,7 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
     case ALGO_YESCRYPTR32:   register_yescryptr32_algo   ( gate ); break;
     case ALGO_YESPOWER:      register_yespower_algo      ( gate ); break;
     case ALGO_YESPOWERR16:   register_yespowerr16_algo   ( gate ); break;
+    case ALGO_YESPOWER_B2B:  register_yespower_b2b_algo  ( gate ); break;
     case ALGO_ZR5:           register_zr5_algo           ( gate ); break;
    default:
       applog(LOG_ERR,"FAIL: algo_gate registration failed, unknown algo %s.\n", algo_names[opt_algo] );
@@ -274,7 +278,7 @@ bool register_json_rpc2( algo_gate_t *gate )
   applog(LOG_WARNING,"supported by cpuminer-opt. Shares submitted will");
   applog(LOG_WARNING,"likely be rejected. Proceed at your own risk.\n");
 
-  gate->wait_for_diff           = (void*)&do_nothing;
+//  gate->wait_for_diff           = (void*)&do_nothing;
   gate->get_new_work            = (void*)&jr2_get_new_work;
   gate->get_nonceptr            = (void*)&jr2_get_nonceptr;
   gate->stratum_gen_work        = (void*)&jr2_stratum_gen_work;
@@ -335,10 +339,10 @@ const char* const algo_alias_map[][2] =
   { "myriad",            "myr-gr"       },
   { "neo",               "neoscrypt"    },
   { "phi",               "phi1612"      },
-//  { "sia",               "blake2b"      },
   { "sib",               "x11gost"      },
   { "timetravel8",       "timetravel"   },
   { "veil",              "x16rt-veil"   },
+  { "x16r-hex",          "hex"          },
   { "yenten",            "yescryptr16"  },
   { "ziftr",             "zr5"          },
   { NULL,                NULL           }   
@@ -362,40 +366,3 @@ void get_algo_alias( char** algo_or_alias )
 #undef ALIAS
 #undef PROPER
 
-bool submit_solution( struct work *work, void *hash,
-                      struct thr_info *thr )
-{
-     work_set_target_ratio( work, hash );
-     if ( submit_work( thr, work ) )
-     {
-         if ( !opt_quiet )
-            applog( LOG_BLUE, "Share %d submitted by thread %d, job %s.",
-                    accepted_share_count + rejected_share_count + 1,
-                    thr->id, work->job_id );
-         return true;
-     }
-     else
-          applog( LOG_WARNING, "Failed to submit share." );
-     return false;
-}
-
-bool submit_lane_solution( struct work *work, void *hash,
-                           struct thr_info *thr, int lane )
-{
-     work_set_target_ratio( work, hash );
-     if ( submit_work( thr, work ) )
-     {
-         if ( !opt_quiet )
-//            applog( LOG_BLUE, "Share %d submitted by thread %d, lane %d.",
-//                    accepted_share_count + rejected_share_count + 1,
-//                    thr->id, lane );
-            applog( LOG_BLUE, "Share %d submitted by thread %d, lane %d, job %s.",
-                    accepted_share_count + rejected_share_count + 1, thr->id,
-                    lane, work->job_id );
-         return true;
-     }
-     else
-          applog( LOG_WARNING, "Failed to submit share." );
-     return false;
-}
-

algo-gate-api.h

@@ -35,7 +35,7 @@
 //    6. Determine if other non existant functions are required.
 //    That is determined by the need to add code in cpu-miner.c
 //    that applies only to the new algo. That is forbidden. All
-//    algo specific code must be in theh algo's file.
+//    algo specific code must be in the algo's file.
 //
 //    7. If new functions need to be added to the gate add the type
 //    to the structure, declare a null instance in this file and define
@@ -48,10 +48,10 @@
 //    instances as they are defined by default, or unsafe functions that
 //    are not needed by the algo.
 //
-//    9. Add an case entry to the switch/case in function register_gate
+//    9. Add a case entry to the switch/case in function register_gate
 //    in file algo-gate-api.c for the new algo.
 //
-//    10 If a new function type was defined add an entry to ini talgo_gate
+//    10 If a new function type was defined add an entry to init algo_gate
 //    to initialize the new function to its null instance described in step 7.
 //
 //    11. If the new algo has aliases add them to the alias array in
@@ -85,14 +85,16 @@
 
 typedef  uint32_t set_t;
 
-#define EMPTY_SET       0
-#define SSE2_OPT        1
-#define AES_OPT         2  
-#define SSE42_OPT       4
-#define AVX_OPT         8
-#define AVX2_OPT     0x10
-#define SHA_OPT      0x20
-#define AVX512_OPT   0x40
+#define EMPTY_SET        0
+#define SSE2_OPT         1
+#define AES_OPT          2  
+#define SSE42_OPT        4
+#define AVX_OPT          8   // Sandybridge
+#define AVX2_OPT      0x10   // Haswell
+#define SHA_OPT       0x20   // sha256 (Ryzen, Ice Lake)
+#define AVX512_OPT    0x40   // AVX512- F, VL, DQ, BW (Skylake-X)
+#define VAES_OPT      0x80   // VAES (Ice Lake)
+
 
 // return set containing all elements from sets a & b
 inline set_t set_union ( set_t a, set_t b ) { return a | b; }
@@ -108,14 +110,7 @@ inline bool set_excl ( set_t a, set_t b ) { return (a & b) == 0; }
 
 typedef struct
 {
-// special case, only one target, provides a callback for scanhash to
-// submit work with less overhead.
-// bool (*submit_work )             ( struct thr_info*, const struct work* );
-
 // mandatory functions, must be overwritten
-// Added a 5th arg for the thread_info structure to replace the int thr id
-// in the first arg. Both will co-exist during the trasition.
-//int ( *scanhash ) ( int, struct work*, uint32_t, uint64_t* );
 int ( *scanhash ) ( struct work*, uint32_t, uint64_t*, struct thr_info* );
 
 // optional unsafe, must be overwritten if algo uses function
@@ -123,28 +118,55 @@ void ( *hash )     ( void*, const void*, uint32_t ) ;
 void ( *hash_suw ) ( void*, const void* );
 
 //optional, safe to use default in most cases
+
+// Allocate thread local buffers and other initialization specific to miner
+// threads.
 bool ( *miner_thread_init )      ( int );
+
+// Generate global blockheader from stratum data.
 void ( *stratum_gen_work )       ( struct stratum_ctx*, struct work* );
+
+// Get thread local copy of blockheader with unique nonce.
 void ( *get_new_work )           ( struct work*, struct work*, int, uint32_t*,
                                    bool );
+
+// Return pointer to nonce in blockheader.
 uint32_t *( *get_nonceptr )      ( uint32_t* );
-void ( *decode_extra_data )      ( struct work*, uint64_t* );
-void ( *wait_for_diff )          ( struct stratum_ctx* );
-int64_t ( *get_max64 )           ();
+
+// Decode getwork blockheader
 bool ( *work_decode )            ( const json_t*, struct work* );
-void ( *set_target)              ( struct work*, double );
+
+// Extra getwork data
+void ( *decode_extra_data )      ( struct work*, uint64_t* );
+
 bool ( *submit_getwork_result )  ( CURL*, struct work* );
+
 void ( *gen_merkle_root )        ( char*, struct stratum_ctx* );
+
+// Increment extranonce
 void ( *build_extraheader )      ( struct work*, struct stratum_ctx* );
+
 void ( *build_block_header )     ( struct work*, uint32_t, uint32_t*,
-	                           uint32_t*, uint32_t, uint32_t );
+	                                uint32_t*, uint32_t, uint32_t );
+// Build mining.submit message
 void ( *build_stratum_request )  ( char*, struct work*, struct stratum_ctx* );
+
 char* ( *malloc_txs_request )    ( struct work* );
+
+// Big or little
 void ( *set_work_data_endian )   ( struct work* );
+
 double ( *calc_network_diff )    ( struct work* );
+
+// Wait for first work
 bool ( *ready_to_mine )          ( struct work*, struct stratum_ctx*, int );
-void ( *resync_threads )         ( struct work* );
+
+// Diverge mining threads
 bool ( *do_this_thread )         ( int );
+
+// After do_this_thread
+void ( *resync_threads )         ( struct work* );
+
 json_t* (*longpoll_rpc_call)     ( CURL*, int*, char* );
 bool ( *stratum_handle_response )( json_t* );
 set_t optimizations;
@@ -193,23 +215,12 @@ void four_way_not_tested();
 // allways returns failure
 int null_scanhash();
 
-// Allow algos to submit from scanhash loop.
-bool submit_solution( struct work *work, void *hash,
-                      struct thr_info *thr );
-bool submit_lane_solution( struct work *work, void *hash,
-                          struct thr_info *thr, int lane );
-
- 
-bool submit_work( struct thr_info *thr, const struct work *work_in );
-
 // displays warning
 void null_hash    ();
 void null_hash_suw();
 
 // optional safe targets, default listed first unless noted.
 
-void std_wait_for_diff();
-
 uint32_t *std_get_nonceptr( uint32_t *work_data );
 uint32_t *jr2_get_nonceptr( uint32_t *work_data );
 
@@ -224,25 +235,13 @@ void jr2_stratum_gen_work( struct stratum_ctx *sctx, struct work *work );
 void sha256d_gen_merkle_root( char *merkle_root, struct stratum_ctx *sctx );
 void SHA256_gen_merkle_root ( char *merkle_root, struct stratum_ctx *sctx );
 
-// pick your favorite or define your own
-int64_t get_max64_0x1fffffLL(); // default
-int64_t get_max64_0x40LL();
-int64_t get_max64_0x3ffff();
-int64_t get_max64_0x3fffffLL();
-int64_t get_max64_0x1ffff();
-int64_t get_max64_0xffffLL();
-
-void std_set_target(    struct work *work, double job_diff );
-void alt_set_target(    struct work* work, double job_diff );
-void scrypt_set_target( struct work *work, double job_diff );
-
 bool std_le_work_decode( const json_t *val, struct work *work );
 bool std_be_work_decode( const json_t *val, struct work *work );
-bool jr2_work_decode( const json_t *val, struct work *work );
+bool jr2_work_decode(    const json_t *val, struct work *work );
 
 bool std_le_submit_getwork_result( CURL *curl, struct work *work );
 bool std_be_submit_getwork_result( CURL *curl, struct work *work );
-bool jr2_submit_getwork_result( CURL *curl, struct work *work );
+bool jr2_submit_getwork_result(    CURL *curl, struct work *work );
 
 void std_le_build_stratum_request( char *req, struct work *work );
 void std_be_build_stratum_request( char *req, struct work *work );
@@ -256,8 +255,8 @@ void set_work_data_big_endian( struct work *work );
 double std_calc_network_diff( struct work *work );
 
 void std_build_block_header( struct work* g_work, uint32_t version,
-	                     uint32_t *prevhash,  uint32_t *merkle_root,
-   	                     uint32_t ntime, uint32_t nbits );
+	                          uint32_t *prevhash,  uint32_t *merkle_root,
+   	                       uint32_t ntime,      uint32_t nbits );
 
 void std_build_extraheader( struct work *work, struct stratum_ctx *sctx );
 
@@ -278,8 +277,8 @@ int std_get_work_data_size();
 // by calling the algo's register function.
 bool register_algo_gate( int algo, algo_gate_t *gate );
 
-// Override any default gate functions that are applicable and do any other
-// algo-specific initialization.
+// Called by algos toverride any default gate functions that are applicable
+// and do any other algo-specific initialization.
 // The register functions for all the algos can be declared here to reduce
 // compiler warnings but that's just more work for devs adding new algos.
 bool register_algo( algo_gate_t *gate );
@@ -292,5 +291,7 @@ bool register_json_rpc2( algo_gate_t *gate );
 // use this to call the hash function of an algo directly, ie util.c test.
 void exec_hash_function( int algo, void *output, const void *pdata );
 
-void get_algo_alias( char** algo_or_alias );
+// Validate a string as a known algo and alias, updates arg to proper
+// algo name if valid alias, NULL if invalid alias or algo.
+void get_algo_alias( char **algo_or_alias );
 

algo/argon2/argon2a/argon2a.c

@@ -74,19 +74,14 @@ int scanhash_argon2( struct work* work, uint32_t max_nonce,
 	return 0;
 }
 
-int64_t argon2_get_max64 ()
-{
-  return 0x1ffLL;
-}
-
 bool register_argon2_algo( algo_gate_t* gate )
 {
   gate->optimizations = SSE2_OPT | AVX_OPT | AVX2_OPT;
   gate->scanhash        = (void*)&scanhash_argon2;
   gate->hash            = (void*)&argon2hash;
   gate->gen_merkle_root = (void*)&SHA256_gen_merkle_root;
-  gate->set_target      = (void*)&scrypt_set_target;
-  gate->get_max64       = (void*)&argon2_get_max64;
+  opt_target_factor = 65536.0;
+
   return true;
 };
 

algo/argon2/argon2d/argon2d-gate.c

@@ -67,8 +67,8 @@ bool register_argon2d_crds_algo( algo_gate_t* gate )
 {
         gate->scanhash = (void*)&scanhash_argon2d_crds;
         gate->hash = (void*)&argon2d_crds_hash;
-        gate->set_target = (void*)&scrypt_set_target;
         gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
+        opt_target_factor = 65536.0;
         return true;
 }
 
@@ -135,8 +135,8 @@ bool register_argon2d_dyn_algo( algo_gate_t* gate )
 {
         gate->scanhash = (void*)&scanhash_argon2d_dyn;
         gate->hash = (void*)&argon2d_dyn_hash;
-        gate->set_target = (void*)&scrypt_set_target;
         gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
+        opt_target_factor = 65536.0;
         return true;
 }
 
@@ -179,14 +179,11 @@ int scanhash_argon2d4096( struct work *work, uint32_t max_nonce,
    return 0;
 }
 
-int64_t get_max64_0x1ff() { return 0x1ff; }
-
 bool register_argon2d4096_algo( algo_gate_t* gate )
 {
         gate->scanhash = (void*)&scanhash_argon2d4096;
-        gate->set_target = (void*)&scrypt_set_target;
-        gate->get_max64  = (void*)&get_max64_0x1ff;
         gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
+        opt_target_factor = 65536.0;
         return true;
 }
 

algo/argon2/argon2d/argon2d/core.c

@@ -28,6 +28,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <mm_malloc.h>
 
 #include "core.h"
 #include "argon2d_thread.h"
@@ -99,7 +100,8 @@ int allocate_memory(const argon2_context *context, uint8_t **memory,
     if (context->allocate_cbk) {
         (context->allocate_cbk)(memory, memory_size);
     } else {
-        *memory = malloc(memory_size);
+        *memory = _mm_malloc( memory_size, 64 );
+//        *memory = malloc(memory_size);
     }
 
     if (*memory == NULL) {
@@ -116,7 +118,8 @@ void free_memory(const argon2_context *context, uint8_t *memory,
     if (context->free_cbk) {
         (context->free_cbk)(memory, memory_size);
     } else {
-        free(memory);
+//        free(memory);
+        _mm_free( memory );
     }
 }
 

algo/argon2/argon2d/argon2d/opt.c

@@ -96,14 +96,14 @@ static void fill_block(__m256i *state, const block *ref_block,
     if (with_xor) {
         for (i = 0; i < ARGON2_HWORDS_IN_BLOCK; i++) {
             state[i] = _mm256_xor_si256(
-                state[i], _mm256_loadu_si256((const __m256i *)ref_block->v + i));
+                state[i], _mm256_load_si256((const __m256i *)ref_block->v + i));
             block_XY[i] = _mm256_xor_si256(
-                state[i], _mm256_loadu_si256((const __m256i *)next_block->v + i));
+                state[i], _mm256_load_si256((const __m256i *)next_block->v + i));
         }
     } else {
         for (i = 0; i < ARGON2_HWORDS_IN_BLOCK; i++) {
             block_XY[i] = state[i] = _mm256_xor_si256(
-                state[i], _mm256_loadu_si256((const __m256i *)ref_block->v + i));
+                state[i], _mm256_load_si256((const __m256i *)ref_block->v + i));
         }
     }
 
@@ -139,7 +139,7 @@ static void fill_block(__m256i *state, const block *ref_block,
 
     for (i = 0; i < ARGON2_HWORDS_IN_BLOCK; i++) {
         state[i] = _mm256_xor_si256(state[i], block_XY[i]);
-        _mm256_storeu_si256((__m256i *)next_block->v + i, state[i]);
+        _mm256_store_si256((__m256i *)next_block->v + i, state[i]);
     }
 }
 

algo/argon2/argon2d/blake2/blamka-round-opt.h

@@ -29,6 +29,8 @@
 #include <x86intrin.h>
 #endif
 
+#include "simd-utils.h"
+
 #if !defined(__AVX512F__)
 #if !defined(__AVX2__)
 #if !defined(__XOP__)
@@ -182,64 +184,63 @@ static BLAKE2_INLINE __m128i fBlaMka(__m128i x, __m128i y) {
 
 #include <immintrin.h>
 
-#define rotr32(x)   _mm256_shuffle_epi32(x, _MM_SHUFFLE(2, 3, 0, 1))
-#define rotr24(x)   _mm256_shuffle_epi8(x, _mm256_setr_epi8(3, 4, 5, 6, 7, 0, 1, 2, 11, 12, 13, 14, 15, 8, 9, 10, 3, 4, 5, 6, 7, 0, 1, 2, 11, 12, 13, 14, 15, 8, 9, 10))
-#define rotr16(x)   _mm256_shuffle_epi8(x, _mm256_setr_epi8(2, 3, 4, 5, 6, 7, 0, 1, 10, 11, 12, 13, 14, 15, 8, 9, 2, 3, 4, 5, 6, 7, 0, 1, 10, 11, 12, 13, 14, 15, 8, 9))
-#define rotr63(x)   _mm256_xor_si256(_mm256_srli_epi64((x), 63), _mm256_add_epi64((x), (x)))
+#define  rotr32  mm256_swap32_64
+#define  rotr24  mm256_ror3x8_64
+#define  rotr16  mm256_ror1x16_64
+#define  rotr63( x ) mm256_rol_64( x, 1 )
+
+//#define rotr32(x)   _mm256_shuffle_epi32(x, _MM_SHUFFLE(2, 3, 0, 1))
+//#define rotr24(x)   _mm256_shuffle_epi8(x, _mm256_setr_epi8(3, 4, 5, 6, 7, 0, 1, 2, 11, 12, 13, 14, 15, 8, 9, 10, 3, 4, 5, 6, 7, 0, 1, 2, 11, 12, 13, 14, 15, 8, 9, 10))
+//#define rotr16(x)   _mm256_shuffle_epi8(x, _mm256_setr_epi8(2, 3, 4, 5, 6, 7, 0, 1, 10, 11, 12, 13, 14, 15, 8, 9, 2, 3, 4, 5, 6, 7, 0, 1, 10, 11, 12, 13, 14, 15, 8, 9))
+//#define rotr63(x)   _mm256_xor_si256(_mm256_srli_epi64((x), 63), _mm256_add_epi64((x), (x)))
 
 #define G1_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
     do { \
-        __m256i ml = _mm256_mul_epu32(A0, B0); \
-        ml = _mm256_add_epi64(ml, ml); \
-        A0 = _mm256_add_epi64(A0, _mm256_add_epi64(B0, ml)); \
+        __m256i ml0, ml1; \
+        ml0 = _mm256_mul_epu32(A0, B0); \
+        ml1 = _mm256_mul_epu32(A1, B1); \
+        ml0 = _mm256_add_epi64(ml0, ml0); \
+        ml1 = _mm256_add_epi64(ml1, ml1); \
+        A0 = _mm256_add_epi64(A0, _mm256_add_epi64(B0, ml0)); \
+        A1 = _mm256_add_epi64(A1, _mm256_add_epi64(B1, ml1)); \
         D0 = _mm256_xor_si256(D0, A0); \
-        D0 = rotr32(D0); \
-        \
-        ml = _mm256_mul_epu32(C0, D0); \
-        ml = _mm256_add_epi64(ml, ml); \
-        C0 = _mm256_add_epi64(C0, _mm256_add_epi64(D0, ml)); \
-        \
-        B0 = _mm256_xor_si256(B0, C0); \
-        B0 = rotr24(B0); \
-        \
-        ml = _mm256_mul_epu32(A1, B1); \
-        ml = _mm256_add_epi64(ml, ml); \
-        A1 = _mm256_add_epi64(A1, _mm256_add_epi64(B1, ml)); \
         D1 = _mm256_xor_si256(D1, A1); \
+        D0 = rotr32(D0); \
         D1 = rotr32(D1); \
-        \
-        ml = _mm256_mul_epu32(C1, D1); \
-        ml = _mm256_add_epi64(ml, ml); \
-        C1 = _mm256_add_epi64(C1, _mm256_add_epi64(D1, ml)); \
-        \
+        ml0 = _mm256_mul_epu32(C0, D0); \
+        ml1 = _mm256_mul_epu32(C1, D1); \
+        ml0 = _mm256_add_epi64(ml0, ml0); \
+        ml1 = _mm256_add_epi64(ml1, ml1); \
+        C0 = _mm256_add_epi64(C0, _mm256_add_epi64(D0, ml0)); \
+        C1 = _mm256_add_epi64(C1, _mm256_add_epi64(D1, ml1)); \
+        B0 = _mm256_xor_si256(B0, C0); \
         B1 = _mm256_xor_si256(B1, C1); \
+        B0 = rotr24(B0); \
         B1 = rotr24(B1); \
     } while((void)0, 0);
 
 #define G2_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
     do { \
-        __m256i ml = _mm256_mul_epu32(A0, B0); \
-        ml = _mm256_add_epi64(ml, ml); \
-        A0 = _mm256_add_epi64(A0, _mm256_add_epi64(B0, ml)); \
+        __m256i ml0, ml1; \
+        ml0 = _mm256_mul_epu32(A0, B0); \
+        ml1 = _mm256_mul_epu32(A1, B1); \
+        ml0 = _mm256_add_epi64(ml0, ml0); \
+        ml1 = _mm256_add_epi64(ml1, ml1); \
+        A0 = _mm256_add_epi64(A0, _mm256_add_epi64(B0, ml0)); \
+        A1 = _mm256_add_epi64(A1, _mm256_add_epi64(B1, ml1)); \
         D0 = _mm256_xor_si256(D0, A0); \
-        D0 = rotr16(D0); \
-        \
-        ml = _mm256_mul_epu32(C0, D0); \
-        ml = _mm256_add_epi64(ml, ml); \
-        C0 = _mm256_add_epi64(C0, _mm256_add_epi64(D0, ml)); \
-        B0 = _mm256_xor_si256(B0, C0); \
-        B0 = rotr63(B0); \
-        \
-        ml = _mm256_mul_epu32(A1, B1); \
-        ml = _mm256_add_epi64(ml, ml); \
-        A1 = _mm256_add_epi64(A1, _mm256_add_epi64(B1, ml)); \
         D1 = _mm256_xor_si256(D1, A1); \
+        D0 = rotr16(D0); \
         D1 = rotr16(D1); \
-        \
-        ml = _mm256_mul_epu32(C1, D1); \
-        ml = _mm256_add_epi64(ml, ml); \
-        C1 = _mm256_add_epi64(C1, _mm256_add_epi64(D1, ml)); \
+        ml0 = _mm256_mul_epu32(C0, D0); \
+        ml1 = _mm256_mul_epu32(C1, D1); \
+        ml0 = _mm256_add_epi64(ml0, ml0); \
+        ml1 = _mm256_add_epi64(ml1, ml1); \
+        C0 = _mm256_add_epi64(C0, _mm256_add_epi64(D0, ml0)); \
+        C1 = _mm256_add_epi64(C1, _mm256_add_epi64(D1, ml1)); \
+        B0 = _mm256_xor_si256(B0, C0); \
         B1 = _mm256_xor_si256(B1, C1); \
+        B0 = rotr63(B0); \
         B1 = rotr63(B1); \
     } while((void)0, 0);
 
@@ -259,16 +260,14 @@ static BLAKE2_INLINE __m128i fBlaMka(__m128i x, __m128i y) {
         __m256i tmp1 = _mm256_blend_epi32(B0, B1, 0xCC); \
         __m256i tmp2 = _mm256_blend_epi32(B0, B1, 0x33); \
         B1 = _mm256_permute4x64_epi64(tmp1, _MM_SHUFFLE(2,3,0,1)); \
-        B0 = _mm256_permute4x64_epi64(tmp2, _MM_SHUFFLE(2,3,0,1)); \
-        \
         tmp1 = C0; \
+        B0 = _mm256_permute4x64_epi64(tmp2, _MM_SHUFFLE(2,3,0,1)); \
         C0 = C1; \
-        C1 = tmp1; \
-        \
-        tmp1 = _mm256_blend_epi32(D0, D1, 0xCC); \
         tmp2 = _mm256_blend_epi32(D0, D1, 0x33); \
-        D0 = _mm256_permute4x64_epi64(tmp1, _MM_SHUFFLE(2,3,0,1)); \
+        C1 = tmp1; \
+        tmp1 = _mm256_blend_epi32(D0, D1, 0xCC); \
         D1 = _mm256_permute4x64_epi64(tmp2, _MM_SHUFFLE(2,3,0,1)); \
+        D0 = _mm256_permute4x64_epi64(tmp1, _MM_SHUFFLE(2,3,0,1)); \
     } while(0);
 
 #define UNDIAGONALIZE_1(A0, B0, C0, D0, A1, B1, C1, D1) \
@@ -287,16 +286,14 @@ static BLAKE2_INLINE __m128i fBlaMka(__m128i x, __m128i y) {
         __m256i tmp1 = _mm256_blend_epi32(B0, B1, 0xCC); \
         __m256i tmp2 = _mm256_blend_epi32(B0, B1, 0x33); \
         B0 = _mm256_permute4x64_epi64(tmp1, _MM_SHUFFLE(2,3,0,1)); \
-        B1 = _mm256_permute4x64_epi64(tmp2, _MM_SHUFFLE(2,3,0,1)); \
-        \
         tmp1 = C0; \
+        B1 = _mm256_permute4x64_epi64(tmp2, _MM_SHUFFLE(2,3,0,1)); \
         C0 = C1; \
-        C1 = tmp1; \
-        \
-        tmp1 = _mm256_blend_epi32(D0, D1, 0x33); \
         tmp2 = _mm256_blend_epi32(D0, D1, 0xCC); \
-        D0 = _mm256_permute4x64_epi64(tmp1, _MM_SHUFFLE(2,3,0,1)); \
+        C1 = tmp1; \
+        tmp1 = _mm256_blend_epi32(D0, D1, 0x33); \
         D1 = _mm256_permute4x64_epi64(tmp2, _MM_SHUFFLE(2,3,0,1)); \
+        D0 = _mm256_permute4x64_epi64(tmp1, _MM_SHUFFLE(2,3,0,1)); \
     } while((void)0, 0);
 
 #define BLAKE2_ROUND_1(A0, A1, B0, B1, C0, C1, D0, D1) \

algo/blake/blake-gate.c

@@ -1,18 +1,8 @@
 #include "blake-gate.h"
 
-int64_t blake_get_max64 ()
-{
-  return 0x7ffffLL;
-}
-
 bool register_blake_algo( algo_gate_t* gate )
 {
   gate->optimizations = AVX2_OPT;
-  gate->get_max64 = (void*)&blake_get_max64;
-//#if defined (__AVX2__) && defined (FOUR_WAY)
-//   gate->optimizations = SSE2_OPT | AVX2_OPT;
-//  gate->scanhash  = (void*)&scanhash_blake_8way;
-//  gate->hash      = (void*)&blakehash_8way;
 #if defined(BLAKE_4WAY)
   four_way_not_tested();
   gate->scanhash  = (void*)&scanhash_blake_4way;

algo/blake/blake256-hash-4way.c

@@ -308,12 +308,12 @@ static const sph_u32 CS[16] = {
 #define GS_4WAY( m0, m1, c0, c1, a, b, c, d ) \
 do { \
    a = _mm_add_epi32( _mm_add_epi32( _mm_xor_si128( \
-                 _mm_set_epi32( c1, c1, c1, c1 ), m0 ), b ), a ); \
+                                   _mm_set1_epi32( c1 ), m0 ), b ), a ); \
    d = mm128_ror_32( _mm_xor_si128( d, a ), 16 ); \
    c = _mm_add_epi32( c, d ); \
    b = mm128_ror_32( _mm_xor_si128( b, c ), 12 ); \
    a = _mm_add_epi32( _mm_add_epi32( _mm_xor_si128( \
-                 _mm_set_epi32( c0, c0, c0, c0 ), m1 ), b ), a ); \
+                                   _mm_set1_epi32( c0 ), m1 ), b ), a ); \
    d = mm128_ror_32( _mm_xor_si128( d, a ), 8 ); \
    c = _mm_add_epi32( c, d ); \
    b = mm128_ror_32( _mm_xor_si128( b, c ), 7 ); \
@@ -508,14 +508,18 @@ do { \
    V5 = H5; \
    V6 = H6; \
    V7 = H7; \
-   V8 = _mm_xor_si128( S0, _mm_set1_epi32( CS0 ) ); \
-   V9 = _mm_xor_si128( S1, _mm_set1_epi32( CS1 ) ); \
-   VA = _mm_xor_si128( S2, _mm_set1_epi32( CS2 ) ); \
-   VB = _mm_xor_si128( S3, _mm_set1_epi32( CS3 ) ); \
-   VC = _mm_xor_si128( _mm_set1_epi32( T0 ), _mm_set1_epi32( CS4 ) ); \
-   VD = _mm_xor_si128( _mm_set1_epi32( T0 ), _mm_set1_epi32( CS5 ) ); \
-   VE = _mm_xor_si128( _mm_set1_epi32( T1 ), _mm_set1_epi32( CS6 ) ); \
-   VF = _mm_xor_si128( _mm_set1_epi32( T1 ), _mm_set1_epi32( CS7 ) ); \
+   V8 = _mm_xor_si128( S0, m128_const1_64( 0x243F6A88243F6A88 ) ); \
+   V9 = _mm_xor_si128( S1, m128_const1_64( 0x85A308D385A308D3 ) ); \
+   VA = _mm_xor_si128( S2, m128_const1_64( 0x13198A2E13198A2E ) ); \
+   VB = _mm_xor_si128( S3, m128_const1_64( 0x0370734403707344 ) ); \
+   VC = _mm_xor_si128( _mm_set1_epi32( T0 ), \
+                           m128_const1_64( 0xA4093822A4093822 ) ); \
+   VD = _mm_xor_si128( _mm_set1_epi32( T0 ), \
+                           m128_const1_64( 0x299F31D0299F31D0 ) ); \
+   VE = _mm_xor_si128( _mm_set1_epi32( T1 ), \
+                           m128_const1_64( 0x082EFA98082EFA98 ) ); \
+   VF = _mm_xor_si128( _mm_set1_epi32( T1 ), \
+                           m128_const1_64( 0xEC4E6C89EC4E6C89 ) ); \
    BLAKE256_4WAY_BLOCK_BSWAP32; \
    ROUND_S_4WAY(0); \
    ROUND_S_4WAY(1); \
@@ -631,16 +635,20 @@ do { \
    V5 = H5; \
    V6 = H6; \
    V7 = H7; \
-   V8 = _mm256_xor_si256( S0, _mm256_set1_epi32( CS0 ) ); \
-   V9 = _mm256_xor_si256( S1, _mm256_set1_epi32( CS1 ) ); \
-   VA = _mm256_xor_si256( S2, _mm256_set1_epi32( CS2 ) ); \
-   VB = _mm256_xor_si256( S3, _mm256_set1_epi32( CS3 ) ); \
-   VC = _mm256_xor_si256( _mm256_set1_epi32( T0 ), _mm256_set1_epi32( CS4 ) ); \
-   VD = _mm256_xor_si256( _mm256_set1_epi32( T0 ), _mm256_set1_epi32( CS5 ) ); \
-   VE = _mm256_xor_si256( _mm256_set1_epi32( T1 ), _mm256_set1_epi32( CS6 ) ); \
-   VF = _mm256_xor_si256( _mm256_set1_epi32( T1 ), _mm256_set1_epi32( CS7 ) ); \
-   shuf_bswap32 = _mm256_set_epi64x( 0x0c0d0e0f08090a0b, 0x0405060700010203, \
-                                     0x0c0d0e0f08090a0b, 0x0405060700010203 ); \
+   V8 = _mm256_xor_si256( S0, m256_const1_64( 0x243F6A88243F6A88 ) ); \
+   V9 = _mm256_xor_si256( S1, m256_const1_64( 0x85A308D385A308D3 ) ); \
+   VA = _mm256_xor_si256( S2, m256_const1_64( 0x13198A2E13198A2E ) ); \
+   VB = _mm256_xor_si256( S3, m256_const1_64( 0x0370734403707344 ) ); \
+   VC = _mm256_xor_si256( _mm256_set1_epi32( T0 ),\
+                              m256_const1_64( 0xA4093822A4093822 ) ); \
+   VD = _mm256_xor_si256( _mm256_set1_epi32( T0 ),\
+                              m256_const1_64( 0x299F31D0299F31D0 ) ); \
+   VE = _mm256_xor_si256( _mm256_set1_epi32( T1 ), \
+                              m256_const1_64( 0x082EFA98082EFA98 ) ); \
+   VF = _mm256_xor_si256( _mm256_set1_epi32( T1 ), \
+                              m256_const1_64( 0xEC4E6C89EC4E6C89 ) ); \
+   shuf_bswap32 = m256_const_64( 0x0c0d0e0f08090a0b, 0x0405060700010203, \
+                                 0x0c0d0e0f08090a0b, 0x0405060700010203 ); \
    M0 = _mm256_shuffle_epi8( * buf    , shuf_bswap32 ); \
    M1 = _mm256_shuffle_epi8( *(buf+ 1), shuf_bswap32 ); \
    M2 = _mm256_shuffle_epi8( *(buf+ 2), shuf_bswap32 ); \
@@ -696,14 +704,14 @@ blake32_4way_init( blake_4way_small_context *ctx, const uint32_t *iv,
                    const uint32_t *salt, int rounds )
 {
    __m128i zero = m128_zero;
-   casti_m128i( ctx->H, 0 ) = _mm_set1_epi32( iv[0] );
-   casti_m128i( ctx->H, 1 ) = _mm_set1_epi32( iv[1] );
-   casti_m128i( ctx->H, 2 ) = _mm_set1_epi32( iv[2] );
-   casti_m128i( ctx->H, 3 ) = _mm_set1_epi32( iv[3] );
-   casti_m128i( ctx->H, 4 ) = _mm_set1_epi32( iv[4] );
-   casti_m128i( ctx->H, 5 ) = _mm_set1_epi32( iv[5] );
-   casti_m128i( ctx->H, 6 ) = _mm_set1_epi32( iv[6] );
-   casti_m128i( ctx->H, 7 ) = _mm_set1_epi32( iv[7] );
+   casti_m128i( ctx->H, 0 ) = m128_const1_64( 0x6A09E6676A09E667 );
+   casti_m128i( ctx->H, 1 ) = m128_const1_64( 0xBB67AE85BB67AE85 );
+   casti_m128i( ctx->H, 2 ) = m128_const1_64( 0x3C6EF3723C6EF372 );
+   casti_m128i( ctx->H, 3 ) = m128_const1_64( 0xA54FF53AA54FF53A );
+   casti_m128i( ctx->H, 4 ) = m128_const1_64( 0x510E527F510E527F );
+   casti_m128i( ctx->H, 5 ) = m128_const1_64( 0x9B05688C9B05688C );
+   casti_m128i( ctx->H, 6 ) = m128_const1_64( 0x1F83D9AB1F83D9AB );
+   casti_m128i( ctx->H, 7 ) = m128_const1_64( 0x5BE0CD195BE0CD19 );
 
    casti_m128i( ctx->S, 0 ) = zero;
    casti_m128i( ctx->S, 1 ) = zero;
@@ -778,12 +786,13 @@ blake32_4way_close( blake_4way_small_context *ctx, unsigned ub, unsigned n,
    else
       ctx->T0 -= 512 - bit_len;
 
-   buf[vptr] = _mm_set1_epi32( 0x80 );
+   buf[vptr] = m128_const1_64( 0x0000008000000080 );
 
    if ( vptr < 12 )
    {
       memset_zero_128( buf + vptr + 1, 13 - vptr  );
-      buf[ 13 ] = _mm_or_si128( buf[ 13 ], _mm_set1_epi32( 0x01000000UL ) );
+      buf[ 13 ] = _mm_or_si128( buf[ 13 ],
+                                m128_const1_64( 0x0100000001000000ULL ) );
       buf[ 14 ] = mm128_bswap_32( _mm_set1_epi32( th ) );
       buf[ 15 ] = mm128_bswap_32( _mm_set1_epi32( tl ) );
       blake32_4way( ctx, buf + vptr, 64 - ptr );
@@ -795,7 +804,8 @@ blake32_4way_close( blake_4way_small_context *ctx, unsigned ub, unsigned n,
       ctx->T0 = 0xFFFFFE00UL;
       ctx->T1 = 0xFFFFFFFFUL;
       memset_zero_128( buf, 56>>2 );
-      buf[ 13 ] = _mm_or_si128( buf[ 13 ], _mm_set1_epi32( 0x01000000UL ) );
+      buf[ 13 ] = _mm_or_si128( buf[ 13 ],
+                                m128_const1_64( 0x0100000001000000ULL ) );
       buf[ 14 ] = mm128_bswap_32( _mm_set1_epi32( th ) );
       buf[ 15 ] = mm128_bswap_32( _mm_set1_epi32( tl ) );
       blake32_4way( ctx, buf, 64 );
@@ -815,20 +825,18 @@ blake32_8way_init( blake_8way_small_context *sc, const sph_u32 *iv,
                    const sph_u32 *salt, int rounds )
 {
    __m256i zero = m256_zero;
-   casti_m256i( sc->H, 0 ) = _mm256_set1_epi32( iv[0] );
-   casti_m256i( sc->H, 1 ) = _mm256_set1_epi32( iv[1] );
-   casti_m256i( sc->H, 2 ) = _mm256_set1_epi32( iv[2] );
-   casti_m256i( sc->H, 3 ) = _mm256_set1_epi32( iv[3] );
-   casti_m256i( sc->H, 4 ) = _mm256_set1_epi32( iv[4] );
-   casti_m256i( sc->H, 5 ) = _mm256_set1_epi32( iv[5] );
-   casti_m256i( sc->H, 6 ) = _mm256_set1_epi32( iv[6] );
-   casti_m256i( sc->H, 7 ) = _mm256_set1_epi32( iv[7] );
-
+   casti_m256i( sc->H, 0 ) = m256_const1_64( 0x6A09E6676A09E667 );
+   casti_m256i( sc->H, 1 ) = m256_const1_64( 0xBB67AE85BB67AE85 );
+   casti_m256i( sc->H, 2 ) = m256_const1_64( 0x3C6EF3723C6EF372 );
+   casti_m256i( sc->H, 3 ) = m256_const1_64( 0xA54FF53AA54FF53A );
+   casti_m256i( sc->H, 4 ) = m256_const1_64( 0x510E527F510E527F );
+   casti_m256i( sc->H, 5 ) = m256_const1_64( 0x9B05688C9B05688C );
+   casti_m256i( sc->H, 6 ) = m256_const1_64( 0x1F83D9AB1F83D9AB );
+   casti_m256i( sc->H, 7 ) = m256_const1_64( 0x5BE0CD195BE0CD19 );
    casti_m256i( sc->S, 0 ) = zero;
    casti_m256i( sc->S, 1 ) = zero;
    casti_m256i( sc->S, 2 ) = zero;
    casti_m256i( sc->S, 3 ) = zero;
-
    sc->T0 = sc->T1 = 0;
    sc->ptr = 0;
    sc->rounds = rounds;
@@ -887,7 +895,7 @@ blake32_8way_close( blake_8way_small_context *sc, unsigned ub, unsigned n,
 
    ptr = sc->ptr;
    bit_len = ((unsigned)ptr << 3);
-   buf[ptr>>2] = _mm256_set1_epi32( 0x80 );
+   buf[ptr>>2] = m256_const1_64( 0x0000008000000080ULL );
    tl = sc->T0 + bit_len;
    th = sc->T1;
 
@@ -909,7 +917,7 @@ blake32_8way_close( blake_8way_small_context *sc, unsigned ub, unsigned n,
        memset_zero_256( buf + (ptr>>2) + 1, (52 - ptr) >> 2 );
        if ( out_size_w32 == 8 )
            buf[52>>2] = _mm256_or_si256( buf[52>>2],
-                                           _mm256_set1_epi32( 0x01000000UL ) );
+                                m256_const1_64( 0x0100000001000000ULL ) );
        *(buf+(56>>2)) = mm256_bswap_32( _mm256_set1_epi32( th ) );
        *(buf+(60>>2)) = mm256_bswap_32( _mm256_set1_epi32( tl ) );
        blake32_8way( sc, buf + (ptr>>2), 64 - ptr );
@@ -922,7 +930,7 @@ blake32_8way_close( blake_8way_small_context *sc, unsigned ub, unsigned n,
         sc->T1 = SPH_C32(0xFFFFFFFFUL);
         memset_zero_256( buf, 56>>2 );
        if ( out_size_w32 == 8 )
-           buf[52>>2] = _mm256_set1_epi32( 0x01000000UL );
+           buf[52>>2] = m256_const1_64( 0x0100000001000000ULL );
         *(buf+(56>>2)) = mm256_bswap_32( _mm256_set1_epi32( th ) );
         *(buf+(60>>2)) = mm256_bswap_32( _mm256_set1_epi32( tl ) );
         blake32_8way( sc, buf, 64 );

algo/blake/blake256-hash-4way.c.new

@@ -1,322 +0,0 @@
-// convert blake256 32 bit to use 64 bit with serial vectoring
-//
-//  cut calls to GS in half
-//
-// combine V
-// v0 = {V0,V1}
-// v1 = {V2,V3}
-// v2 = {V4,V5}
-// v3 = {V6,V7}
-// v4 = {V8,V9}
-// v5 = {VA,VB}
-// v6 = {VC,VD}
-// v7 = {CE,VF}
-//
-// v6x = {VD,VC}      swap(VC,VD)   swap(v6)
-// v7x = {VF,VE}      swap(VE,VF)   swap(v7)
-//
-// V0 = v1v0
-// V1 = v3v2
-// V2 = v5v4
-// V3 = v7v6
-// V4 = v9v8
-// V5 = vbva
-// V6 = vdvc
-// V7 = vfve
-//
-// The rotate in ROUND is to effect straddle and unstraddle for the third
-// and 4th iteration of GS.
-// It concatenates 2 contiguous 256 bit vectors and extracts the middle
-// 256 bits. After the transform they must be restored with only the
-// chosen bits modified in the original 2 vectors.
-// ror1x128 achieves this by putting the chosen bits in arg1, the "low"
-// 256 bit vector and saves the untouched bits temporailly in arg0, the
-// "high" 256 bit vector. Simply reverse the process to restore data back
-// to original positions.
-
-// Use standard 4way when AVX2 is not available use x2 mode with AVX2.
-//
-// Data is organised the same as 32 bit 4 way, in effect serial vectoring
-// on top of parallel vectoring. Same data in the same place just taking
-// two chunks at a time.
-//
-// Transparent to user, x2 mode used when AVX2 detected.
-// Use existing 4way context but revert to scalar types.
-// Same interleave function (128 bit) or x2 with 256 bit?
-// User trsnaparency would have to apply to interleave as well.
-//
-// Use common 4way update and close
-
-/*
-typedef struct {
-   unsigned char buf[64<<2];
-   uint32_t H[8<<2];
-   uint32_t S[4<<2];
-   size_t ptr;
-   uint32_t T0, T1;
-   int rounds;   // 14 for blake, 8 for blakecoin & vanilla
-} blakex2_4way_small_context __attribute__ ((aligned (64)));
-*/
-
-static void
-blake32x2_4way_init( blake_4way_small_context *ctx, const uint32_t *iv,
-                   const uint32_t *salt, int rounds )
-{
-   casti_m128i( ctx->H, 0 ) = _mm_set1_epi32( iv[0] );
-   casti_m128i( ctx->H, 1 ) = _mm_set1_epi32( iv[1] );
-   casti_m128i( ctx->H, 2 ) = _mm_set1_epi32( iv[2] );
-   casti_m128i( ctx->H, 3 ) = _mm_set1_epi32( iv[3] );
-   casti_m128i( ctx->H, 4 ) = _mm_set1_epi32( iv[4] );
-   casti_m128i( ctx->H, 5 ) = _mm_set1_epi32( iv[5] );
-   casti_m128i( ctx->H, 6 ) = _mm_set1_epi32( iv[6] );
-   casti_m128i( ctx->H, 7 ) = _mm_set1_epi32( iv[7] );
-
-   casti_m128i( ctx->S, 0 ) = m128_zero;
-   casti_m128i( ctx->S, 1 ) = m128_zero;
-   casti_m128i( ctx->S, 2 ) = m128_zero;
-   casti_m128i( ctx->S, 3 ) = m128_zero;
-/*
-   sc->S[0] = _mm_set1_epi32( salt[0] );
-   sc->S[1] = _mm_set1_epi32( salt[1] );
-   sc->S[2] = _mm_set1_epi32( salt[2] );
-   sc->S[3] = _mm_set1_epi32( salt[3] );
-*/
-   ctx->T0 = ctx->T1 = 0;
-   ctx->ptr = 0;
-   ctx->rounds = rounds;
-}
-
-static void
-blake32x2( blake_4way_small_context *ctx, const void *data, size_t len )
-{
-   __m128i *buf = (__m256i*)ctx->buf;
-   size_t  bptr = ctx->ptr << 2;
-   size_t  vptr = ctx->ptr >> 3;
-   size_t  blen = len << 2;
-//    unsigned char *buf = ctx->buf;
-//    size_t ptr         = ctx->ptr<<4;  // repurposed
-    DECL_STATE32x2
-
-//    buf = sc->buf;
-//    ptr = sc->ptr;
-
-// adjust len for use with ptr, clen, all absolute bytes.
-//    int blen = len<<2;
-
-    if ( blen < (sizeof ctx->buf) - bptr )
-    {
-        memcpy( buf + vptr, data, blen );
-        ptr += blen;
-        ctx->ptr = bptr >> 2;;
-        return;
-    }
-
-    READ_STATE32( ctx );
-    while ( blen > 0 )
-    {
-        size_t clen;
-
-        clen = ( sizeof sc->buf ) - ptr;
-        if ( clen > blen )
-            clen = blen;
-        memcpy( buf + vptr, data, clen );
-        bptr += clen;
-        vptr = bptr >> 5;
-	data = (const unsigned char *)data + clen;
-        blen -= clen;
-        if ( bptr == sizeof ctx->buf )
-       	{
-           if ( ( T0 = T0 + 512 ) < 512 ) // not needed, will never rollover
-               T1 += 1;
-           COMPRESS32x2_4WAY( ctx->rounds );
-           ptr = 0;
-        }
-    }
-    WRITE_STATE32x2( ctx );
-    ctx->ptr = bptr >> 2;
-}
-
-static void
-blake32x2_4way_close( blake_4way_small_context *ctx, void *dst )
-{
-   __m256i buf[8] __attribute__ ((aligned (64)));
-   size_t   ptr     = ctx->ptr;
-   size_t   vptr    = ctx->ptr>>2;
-   unsigned bit_len = ( (unsigned)ptr << 3 );  // one lane
-   uint32_t th      = ctx->T1;
-   uint32_t tl      = ctx->T0 + bit_len;
-
-   if ( ptr == 0 )
-   {
-        ctx->T0 = 0xFFFFFE00UL;
-        ctx->T1 = 0xFFFFFFFFUL;
-   }
-   else if ( ctx->T0 == 0 )
-   {
-      ctx->T0 = 0xFFFFFE00UL + bit_len;
-      ctx->T1 -= 1;
-   }
-   else
-      ctx->T0 -= 512 - bit_len;
-
-   // memset doesn't do ints
-   buf[ vptr ] = _mm256_set_epi32( 0,0,0,0, 0x80, 0x80, 0x80, 0x80 );
-
-   if ( vptr < 5 )
-   {
-       memset_zero_256( buf + vptr + 1, 6 - vptr  );
-       buf[ 6 ] = _mm256_or_si256( vbuf[ 6 ], _mm256_set_epi32(
-             0x01000000UL,0x01000000UL,0x01000000UL,0x01000000UL, 0,0,0,0 ) ); 
-       buf[ 7 ] = mm256_bswap_32( _mm256_set_epi32( tl,tl,tl,tl,
-			                            th,th,th,th ) );
-       blake32x2_4way( ctx, buf + vptr, 64 - ptr );
-   }
-   else
-   {
-       memset_zero_256( vbuf + vptr + 1, 7 - vptr );
-       blake32x2_4way( ctx,  vbuf + ptr, 64 - ptr );
-       ctx->T0 = 0xFFFFFE00UL;
-       ctx->T1 = 0xFFFFFFFFUL;
-       buf[ 6 ] = mm256_zero;
-       buf[ 6 ] = _mm256_set_epi32( 0,0,0,0,
-		         0x01000000UL,0x01000000UL,0x01000000UL,0x01000000UL );
-       buf[ 7 ] = mm256_bswap_32( _mm256_set_epi32( tl, tl, tl, tl,
-                                                    th, th, th, th );
-       blake32x2_4way( ctx, buf, 64 );
-   }
-
-   casti_m256i( dst, 0 ) = mm256_bswap_32( casti_m256i( ctx->H, 0 ) );
-   casti_m256i( dst, 1 ) = mm256_bswap_32( casti_m256i( ctx->H, 1 ) );
-   casti_m256i( dst, 2 ) = mm256_bswap_32( casti_m256i( ctx->H, 2 ) );
-   casti_m256i( dst, 3 ) = mm256_bswap_32( casti_m256i( ctx->H, 3 ) );
-}
-
-
-
-
-#define DECL_STATE32x2_4WAY \
-   __m256i H0, H1, H2, H3; \
-   __m256i S0, S1; \
-   uint32_t T0, T1;
-
-#define READ_STATE32x2_4WAY(state)  do \
-{ \
-   H0 = casti_m256i( state->H, 0 ); \
-   H1 = casti_m256i( state->H, 1 ); \
-   H2 = casti_m256i( state->H, 2 ); \
-   H3 = casti_m256i( state->H, 3 ); \
-   S0 = casti_m256i( state->S, 0 ); \
-   S1 = casti_m256i( state->S, 1 ); \
-   T0 = state->T0; \
-   T1 = state->T1; \
-
-#define WRITE_STATE32x2_4WAY(state)   do { \
-   casti_m256i( state->H, 0 ) = H0; \
-   casti_m256i( state->H, 1 ) = H1; \
-   casti_m256i( state->H, 2 ) = H2; \
-   casti_m256i( state->H, 3 ) = H3; \
-   casti_m256i( state->S, 0 ) = S0; \
-   casti_m256i( state->S, 1 ) = S1; \
-   state->T0 = T0; \
-   state->T1 = T1; \
-} while (0)
-
-
-#define GSx2_4WAY( m0m2, m1m3, c0c2, c1c3, a, b, c, d ) do \
-{ \
-   a = _mm256_add_epi32( _mm256_add_epi32( _mm256_xor_si256( \
-          _mm256_set_epi32( c1,c3, c1,c3, c1,c3, c1,c3 ), \
-	  _mm256_set_epi32( m0,m2, m0,m2, m0,m2, m0,m2 ) ), b ), a ); \
-   d = mm256_ror_32( _mm_xor_si128( d, a ), 16 ); \
-   c = _mm256_add_epi32( c, d ); \
-   b = mm256_ror_32( _mm256_xor_si256( b, c ), 12 ); \
-   a = _mm256_add_epi32( _mm256_add_epi32( _mm256_xor_si256( \
-          _mm256_set_epi32( c0,c2, c0,c2, c0,c2, c0,c2 ), \
-	  _mm256_set_epi32( m1,m3, m1,m3, m1,m3, m1,m3 ) ), b ), a ); \
-   d = mm256_ror_32( _mm256_xor_si256( d, a ), 8 ); \
-   c = _mm256_add_epi32( c, d ); \
-   b = mm256_ror_32( _mm256_xor_si256( b, c ), 7 ); \
-} while (0)
-
-#define ROUND_Sx2_4WAY(r)   do \
-{ \
-  GS2_4WAY( Mx(r, 0),  Mx(r, 1),  Mx(r, 2),  Mx(r, 3), \
-           CSx(r, 0), CSx(r, 1), CSx(r, 2), CSx(r, 3), V0, V2, V4, V6 ); \
-  GS2_4WAY( Mx(r, 4),  Mx(r, 5),  Mx(r, 6),  Mx(r, 7), \
-           CSx(r, 4), CSx(r, 5), CSx(r, 6), CSx(r, 7), V1, V3, V5, V7 ); \
-  mm256_ror1x128_512( V3, V2 ); \
-  mm256_ror1x128_512( V6, V7 ); \
-  GS2_4WAY( Mx(r, 8),  Mx(r, 9),  Mx(r, A),  Mx(r, B), \
-           CSx(r, 8), CSx(r, 9), CSx(r, A), CSx(r, B), V0, V2, V5, V7 ); \
-  GS2_4WAY( Mx(r, C),  Mx(r, D),  Mx(r, C),  Mx(r, D), \
-           CSx(r, C), CSx(r, D), CSx(r, C), CSx(r, D), V1, V3, V4, V6 ); \
-  mm256_rol1x128_512( V2, V3 ); \
-  mm256_rol1x128_512( V7, V6 ); 
-
-#define COMPRESS32x2_4WAY( rounds ) do \
-{ \
-   __m256i M0, M1, M2, M3, M4, M5, M6, M7; \
-   __m256i V0, V1, V2, V3, V4, V5, V6, V7; \
-   unsigned r; \
-   V0 = H0; \
-   V1 = H1; \
-   V2 = H2; \
-   V3 = H3; \
-   V4 = _mm256_xor_si256( S0, _mm256_set_epi32( CS1, CS1, CS1, CS1, \
-			                        CS0, CS0, CS0, CS0 ) ); \
-   V5 = _mm256_xor_si256( S1, _mm256_set_epi32( CS3, CS3, CS3, CS3, \
-                                                CS2, CS2, CS2, CS2 ) ); \
-   V6 = _mm256_xor_si256( _mm256_set1_epi32( T0 ), \
-                              _mm256_set_epi32( CS5, CS5, CS5, CS5, \
-		                                CS4, CS4, CS4, CS4 ) ); \
-   V7 = _mm256_xor_si256( _mm256_set1_epi32( T1 ), \
-                              _mm256_set_epi32( CS7, CS7, CS7, CS7, \
-                                                CS6, CS6, CS6, CS6 ) ); \
-   M0 = mm256_bswap_32( buf[ 0] ); \
-   M1 = mm256_bswap_32( buf[ 1] ); \
-   M2 = mm256_bswap_32( buf[ 2] ); \
-   M3 = mm256_bswap_32( buf[ 3] ); \
-   M4 = mm256_bswap_32( buf[ 4] ); \
-   M5 = mm256_bswap_32( buf[ 5] ); \
-   M6 = mm256_bswap_32( buf[ 6] ); \
-   M7 = mm256_bswap_32( buf[ 7] ); \
-   ROUND_Sx2_4WAY(0); \
-   ROUND_Sx2_4WAY(1); \
-   ROUND_Sx2_4WAY(2); \
-   ROUND_Sx2_4WAY(3); \
-   ROUND_Sx2_4WAY(4); \
-   ROUND_Sx2_4WAY(5); \
-   ROUND_Sx2_4WAY(6); \
-   ROUND_Sx2_4WAY(7); \
-   if (rounds == 14) \
-   { \
-      ROUND_Sx2_4WAY(8); \
-      ROUND_Sx2_4WAY(9); \
-      ROUND_Sx2_4WAY(0); \
-      ROUND_Sx2_4WAY(1); \
-      ROUND_Sx2_4WAY(2); \
-      ROUND_Sx2_4WAY(3); \
-   } \
-   H0 = _mm256_xor_si256( _mm256_xor_si256( \
-			           _mm256_xor_si256( V8, V0 ), S0 ), H0 ); \
-   H1 = _mm256_xor_si256( _mm256_xor_si256( \
-			           _mm256_xor_si256( V9, V1 ), S1 ), H1 ); \
-   H2 = _mm256_xor_si256( _mm256_xor_si256( \
-			           _mm256_xor_si256( VA, V2 ), S2 ), H2 ); \
-   H3 = _mm256_xor_si256( _mm256_xor_si256( \
-			           _mm256_xor_si256( VB, V3 ), S3 ), H3 ); \
-} while (0)
-
-
-
-
-
-
-
-
-
-
-
-
-
-

algo/blake/blake2b-4way.c

@@ -0,0 +1,67 @@
+/**
+ * Blake2-B Implementation
+ * tpruvot@github 2015-2016
+ */
+
+#include "blake2b-gate.h"
+
+#if defined(BLAKE2B_4WAY)
+
+#include <string.h>
+#include <stdint.h>
+#include "blake2b-hash-4way.h"
+
+// Function not used, code inlined.
+void blake2b_4way_hash(void *output, const void *input)
+{
+    blake2b_4way_ctx ctx;
+    blake2b_4way_init( &ctx );
+    blake2b_4way_update( &ctx, input, 80 );
+    blake2b_4way_final( &ctx, output );
+}
+
+int scanhash_blake2b_4way( struct work *work, uint32_t max_nonce,
+                           uint64_t *hashes_done, struct thr_info *mythr )
+{
+	uint32_t hash[8*4] __attribute__ ((aligned (64)));;
+   uint32_t vdata[20*4] __attribute__ ((aligned (32)));;
+   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
+   blake2b_4way_ctx ctx __attribute__ ((aligned (32)));
+   uint32_t *hash7 = &(hash[25]);   // 3*8+1
+	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];
+	const uint32_t first_nonce = pdata[19];
+
+	uint32_t n = first_nonce;
+
+   mm256_bswap32_intrlv80_4x64( vdata, pdata );
+
+	do {
+      *noncev = mm256_intrlv_blend_32( mm256_bswap_32(
+                _mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ) ), *noncev );
+
+      blake2b_4way_init( &ctx ); 
+      blake2b_4way_update( &ctx, vdata, 80 );
+      blake2b_4way_final( &ctx, hash );
+
+      for ( int lane = 0; lane < 4; lane++ )
+      if ( hash7[ lane<<1 ] < Htarg )
+      {
+          extr_lane_4x64( 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 += 4;
+   } while ( (n < max_nonce-4) && !work_restart[thr_id].restart);
+
+   *hashes_done = n - first_nonce + 1;
+   return 0;
+}
+
+#endif

algo/blake/blake2b-gate.c

@@ -0,0 +1,16 @@
+#include "blake2b-gate.h"
+
+bool register_blake2b_algo( algo_gate_t* gate )
+{
+#if defined(BLAKE2B_4WAY)
+  gate->scanhash  = (void*)&scanhash_blake2b_4way;
+  gate->hash      = (void*)&blake2b_4way_hash;
+#else
+  gate->scanhash  = (void*)&scanhash_blake2b;
+  gate->hash      = (void*)&blake2b_hash;
+#endif
+  gate->optimizations =  AVX2_OPT;
+  return true;
+};
+
+

algo/blake/blake2b-gate.h

@@ -0,0 +1,26 @@
+#ifndef __BLAKE2B_GATE_H__
+#define __BLAKE2B_GATE_H__ 1
+
+#include <stdint.h>
+#include "algo-gate-api.h"
+
+#if defined(__AVX2__)
+  #define BLAKE2B_4WAY
+#endif
+
+bool register_blake2b_algo( algo_gate_t* gate );
+
+#if defined(BLAKE2B_4WAY)
+
+void blake2b_4way_hash( void *state, const void *input );
+int scanhash_blake2b_4way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
+#else
+
+void blake2b_hash( void *state, const void *input );
+int scanhash_blake2b( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr );
+
+#endif
+
+#endif

algo/blake/blake2b-hash-4way.c

@@ -0,0 +1,215 @@
+/*
+ * Copyright 2009 Colin Percival, 2014 savale
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * This file was originally written by Colin Percival as part of the Tarsnap
+ * online backup system.
+ */
+
+#include <stdlib.h>
+#include <stdint.h>
+#include <string.h>
+
+#include "blake2b-hash-4way.h"
+
+#if defined(__AVX2__)
+
+// G Mixing function.
+
+#define B2B_G(a, b, c, d, x, y) \
+{ \
+   v[a] = _mm256_add_epi64( _mm256_add_epi64( v[a], v[b] ), x ); \
+	v[d] = mm256_ror_64( _mm256_xor_si256( v[d], v[a] ), 32 ); \
+	v[c] = _mm256_add_epi64( v[c], v[d] ); \
+	v[b] = mm256_ror_64( _mm256_xor_si256( v[b], v[c] ), 24 ); \
+	v[a] = _mm256_add_epi64( _mm256_add_epi64( v[a], v[b] ), y ); \
+	v[d] = mm256_ror_64( _mm256_xor_si256( v[d], v[a] ), 16 ); \
+	v[c] = _mm256_add_epi64( v[c], v[d] ); \
+	v[b] = mm256_ror_64( _mm256_xor_si256( v[b], v[c] ), 63 ); \
+}
+
+// Initialization Vector.
+/*
+static const uint64_t blake2b_iv[8] = {
+	0x6A09E667F3BCC908, 0xBB67AE8584CAA73B,
+	0x3C6EF372FE94F82B, 0xA54FF53A5F1D36F1,
+	0x510E527FADE682D1, 0x9B05688C2B3E6C1F,
+	0x1F83D9ABFB41BD6B, 0x5BE0CD19137E2179
+};
+*/
+
+static void blake2b_4way_compress( blake2b_4way_ctx *ctx, int last )
+{
+	const uint8_t sigma[12][16] = {
+		{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
+		{ 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 },
+		{ 11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4 },
+		{ 7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8 },
+		{ 9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13 },
+		{ 2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9 },
+		{ 12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11 },
+		{ 13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10 },
+		{ 6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5 },
+		{ 10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13, 0 },
+		{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
+		{ 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 }
+	};
+	int i;
+	__m256i v[16], m[16];
+
+   v[ 0] = ctx->h[0];
+   v[ 1] = ctx->h[1];
+   v[ 2] = ctx->h[2];
+   v[ 3] = ctx->h[3];
+   v[ 4] = ctx->h[4];
+   v[ 5] = ctx->h[5];
+   v[ 6] = ctx->h[6];
+   v[ 7] = ctx->h[7];
+   v[ 8] = m256_const1_64( 0x6A09E667F3BCC908 );
+   v[ 9] = m256_const1_64( 0xBB67AE8584CAA73B );
+   v[10] = m256_const1_64( 0x3C6EF372FE94F82B );
+   v[11] = m256_const1_64( 0xA54FF53A5F1D36F1 );
+   v[12] = m256_const1_64( 0x510E527FADE682D1 );
+   v[13] = m256_const1_64( 0x9B05688C2B3E6C1F );
+   v[14] = m256_const1_64( 0x1F83D9ABFB41BD6B );
+   v[15] = m256_const1_64( 0x5BE0CD19137E2179 );
+
+   v[12] = _mm256_xor_si256( v[12], _mm256_set1_epi64x( ctx->t[0] ) );
+   v[13] = _mm256_xor_si256( v[13], _mm256_set1_epi64x( ctx->t[1] ) );
+
+   if ( last )   
+		v[14] = mm256_not( v[14] );
+
+   m[ 0] = ctx->b[ 0];
+   m[ 1] = ctx->b[ 1];
+   m[ 2] = ctx->b[ 2];
+   m[ 3] = ctx->b[ 3];
+   m[ 4] = ctx->b[ 4];
+   m[ 5] = ctx->b[ 5];
+   m[ 6] = ctx->b[ 6];
+   m[ 7] = ctx->b[ 7];
+   m[ 8] = ctx->b[ 8];
+   m[ 9] = ctx->b[ 9];
+   m[10] = ctx->b[10];
+   m[11] = ctx->b[11];
+   m[12] = ctx->b[12];
+   m[13] = ctx->b[13];
+   m[14] = ctx->b[14];
+   m[15] = ctx->b[15];
+   
+	for ( i = 0; i < 12; i++ )
+   { 
+		B2B_G( 0, 4,  8, 12, m[ sigma[i][ 0] ], m[ sigma[i][ 1] ] );
+		B2B_G( 1, 5,  9, 13, m[ sigma[i][ 2] ], m[ sigma[i][ 3] ] );
+		B2B_G( 2, 6, 10, 14, m[ sigma[i][ 4] ], m[ sigma[i][ 5] ] );
+		B2B_G( 3, 7, 11, 15, m[ sigma[i][ 6] ], m[ sigma[i][ 7] ] );
+		B2B_G( 0, 5, 10, 15, m[ sigma[i][ 8] ], m[ sigma[i][ 9] ] );
+		B2B_G( 1, 6, 11, 12, m[ sigma[i][10] ], m[ sigma[i][11] ] );
+		B2B_G( 2, 7,  8, 13, m[ sigma[i][12] ], m[ sigma[i][13] ] );
+		B2B_G( 3, 4,  9, 14, m[ sigma[i][14] ], m[ sigma[i][15] ] );
+	}
+
+   ctx->h[0] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[0], v[0] ), v[ 8] );
+   ctx->h[1] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[1], v[1] ), v[ 9] );
+   ctx->h[2] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[2], v[2] ), v[10] );
+   ctx->h[3] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[3], v[3] ), v[11] );
+   ctx->h[4] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[4], v[4] ), v[12] );
+   ctx->h[5] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[5], v[5] ), v[13] );
+   ctx->h[6] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[6], v[6] ), v[14] );
+   ctx->h[7] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[7], v[7] ), v[15] );
+}
+
+int blake2b_4way_init( blake2b_4way_ctx *ctx ) 
+{
+	size_t i;
+
+   ctx->h[0] = m256_const1_64( 0x6A09E667F3BCC908 );
+   ctx->h[1] = m256_const1_64( 0xBB67AE8584CAA73B );
+   ctx->h[2] = m256_const1_64( 0x3C6EF372FE94F82B );
+   ctx->h[3] = m256_const1_64( 0xA54FF53A5F1D36F1 );
+   ctx->h[4] = m256_const1_64( 0x510E527FADE682D1 );
+   ctx->h[5] = m256_const1_64( 0x9B05688C2B3E6C1F );
+   ctx->h[6] = m256_const1_64( 0x1F83D9ABFB41BD6B );
+   ctx->h[7] = m256_const1_64( 0x5BE0CD19137E2179 );
+
+   ctx->h[0] = _mm256_xor_si256( ctx->h[0], m256_const1_64( 0x01010020 ) );
+
+	ctx->t[0] = 0;
+	ctx->t[1] = 0;
+	ctx->c = 0;
+	ctx->outlen = 32;
+
+   for ( i = 0; i < 16; i++ )
+     ctx->b[i] = m256_zero;
+
+	return 0;
+}
+
+void blake2b_4way_update( blake2b_4way_ctx *ctx, const void *input,
+                          size_t inlen ) 
+{
+   __m256i* in =(__m256i*)input;
+
+	size_t i, c;
+   c = ctx->c >> 3; 
+
+	for ( i = 0; i < (inlen >> 3); i++ )
+   {
+		if ( ctx->c == 128 )
+      { 
+			ctx->t[0] += ctx->c;
+			if ( ctx->t[0] < ctx->c )
+				ctx->t[1]++;
+			blake2b_4way_compress( ctx, 0 );
+			ctx->c = 0;
+		}
+      ctx->b[ c++ ] = in[i];
+      ctx->c += 8;
+   }
+}
+
+void blake2b_4way_final( blake2b_4way_ctx *ctx, void *out )
+{
+	size_t c;
+   c = ctx->c >> 3;
+
+	ctx->t[0] += ctx->c;
+	if ( ctx->t[0] < ctx->c )
+		ctx->t[1]++;
+
+	while ( ctx->c < 128 )
+   {
+      ctx->b[c++] = m256_zero;
+      ctx->c += 8;
+   }
+
+   blake2b_4way_compress( ctx, 1 );           // final block flag = 1
+
+   casti_m256i( out, 0 ) = ctx->h[0];
+   casti_m256i( out, 1 ) = ctx->h[1];
+   casti_m256i( out, 2 ) = ctx->h[2];
+   casti_m256i( out, 3 ) = ctx->h[3];
+}
+
+#endif

algo/blake/blake2b-hash-4way.h

@@ -0,0 +1,35 @@
+#pragma once
+#ifndef __BLAKE2B_HASH_4WAY_H__
+#define __BLAKE2B_HASH_4WAY_H__
+
+#if defined(__AVX2__)
+
+#include "simd-utils.h"
+#include <stddef.h>
+#include <stdint.h>
+
+#if defined(_MSC_VER)
+#include <inttypes.h>
+#define inline __inline
+#define ALIGN(x) __declspec(align(x))
+#else
+#define ALIGN(x) __attribute__((aligned(x)))
+#endif
+
+// state context
+ALIGN(64) typedef struct {
+	__m256i b[16]; // input buffer
+	__m256i h[8];  // chained state
+	uint64_t t[2];  // total number of bytes
+	size_t c;       // pointer for b[]
+	size_t outlen;  // digest size
+} blake2b_4way_ctx __attribute__((aligned(64)));
+
+int blake2b_4way_init( blake2b_4way_ctx *ctx );
+void blake2b_4way_update( blake2b_4way_ctx *ctx, const void *input,
+                          size_t inlen );
+void blake2b_4way_final( blake2b_4way_ctx *ctx, void *out );
+
+#endif
+
+#endif

algo/blake/blake2b.c

@@ -3,13 +3,11 @@
  * tpruvot@github 2015-2016
  */
 
-#include "algo-gate-api.h"
+#include "blake2b-gate.h"
 #include <string.h>
 #include <stdint.h>
 #include "algo/blake/sph_blake2b.h"
 
-//static __thread sph_blake2b_ctx s_midstate;
-//static __thread sph_blake2b_ctx s_ctx;
 #define MIDLEN 76
 #define A 64
 
@@ -25,16 +23,6 @@ void blake2b_hash(void *output, const void *input)
 	memcpy(output, hash, 32);
 }
 
-/*
-static void blake2b_hash_end(uint32_t *output, const uint32_t *input)
-{
-	s_ctx.outlen = MIDLEN;
-	memcpy(&s_ctx, &s_midstate, 32 + 16 + MIDLEN);
-	sph_blake2b_update(&s_ctx, (uint8_t*) &input[MIDLEN/4], 80 - MIDLEN);
-	sph_blake2b_final(&s_ctx, (uint8_t*) output);
-}
-*/
-
 int scanhash_blake2b( struct work *work, uint32_t max_nonce,
                       uint64_t *hashes_done, struct thr_info *mythr )
 {
@@ -45,7 +33,7 @@ int scanhash_blake2b( struct work *work, uint32_t max_nonce,
    int thr_id = mythr->id;  // thr_id arg is deprecated
 
 	const uint32_t Htarg = ptarget[7];
-	const uint32_t first_nonce = pdata[8];
+	const uint32_t first_nonce = pdata[19];
 
 	uint32_t n = first_nonce;
 
@@ -53,179 +41,23 @@ int scanhash_blake2b( struct work *work, uint32_t max_nonce,
 		be32enc(&endiandata[i], pdata[i]);
 	}
 
-	// midstate (untested yet)
-	//blake2b_init(&s_midstate, 32, NULL, 0);
-	//blake2b_update(&s_midstate, (uint8_t*) endiandata, MIDLEN);
-	//memcpy(&s_ctx, &s_midstate, sizeof(blake2b_ctx));
-
 	do {
-		be32enc(&endiandata[8], n);
+		be32enc(&endiandata[19], n);
 		//blake2b_hash_end(vhashcpu, endiandata);
 		blake2b_hash(vhashcpu, endiandata);
 
 		if (vhashcpu[7] < Htarg && fulltest(vhashcpu, ptarget)) {
 			work_set_target_ratio(work, vhashcpu);
 			*hashes_done = n - first_nonce + 1;
-			pdata[8] = n;
+			pdata[19] = n;
 			return 1;
 		}
 		n++;
 
 	} while (n < max_nonce && !work_restart[thr_id].restart);
 	*hashes_done = n - first_nonce + 1;
-	pdata[8] = n;
+	pdata[19] = n;
 
 	return 0;
 }
 
-static inline void swab256(void *dest_p, const void *src_p)
-{
-	uint32_t *dest = (uint32_t *)dest_p;
-	const uint32_t *src = (uint32_t *)src_p;
-
-	dest[0] = swab32(src[7]);
-	dest[1] = swab32(src[6]);
-	dest[2] = swab32(src[5]);
-	dest[3] = swab32(src[4]);
-	dest[4] = swab32(src[3]);
-	dest[5] = swab32(src[2]);
-	dest[6] = swab32(src[1]);
-	dest[7] = swab32(src[0]);
-}
-
-/* compute nbits to get the network diff */
-void blake2b_calc_network_diff(struct work *work)
-{
-        // sample for diff 43.281 : 1c05ea29
-        uint32_t nbits = work->data[11]; // unsure if correct
-        uint32_t bits = (nbits & 0xffffff);
-        int16_t shift = (swab32(nbits) & 0xff); // 0x1c = 28
-
-        double d = (double)0x0000ffff / (double)bits;
-        for (int m=shift; m < 29; m++) d *= 256.0;
-        for (int m=29; m < shift; m++) d /= 256.0;
-        if (opt_debug_diff)
-                applog(LOG_DEBUG, "net diff: %f -> shift %u, bits %08x", d, shift, bits);
-        net_diff = d;
-}
-
-void blake2b_be_build_stratum_request( char *req, struct work *work )
-{
-   unsigned char *xnonce2str;
-   uint32_t ntime,       nonce;
-   char     ntimestr[9], noncestr[9];
-   be32enc( &ntime, work->data[ algo_gate.ntime_index ] );
-   be32enc( &nonce, work->data[ algo_gate.nonce_index ] );
-   bin2hex( ntimestr, (char*)(&ntime), sizeof(uint32_t) );
-   bin2hex( noncestr, (char*)(&nonce), sizeof(uint32_t) );
-   uint16_t high_nonce = swab32(work->data[9]) >> 16;
-   xnonce2str = abin2hex((unsigned char*)(&high_nonce), 2);
-   snprintf( req, JSON_BUF_LEN,
-        "{\"method\": \"mining.submit\", \"params\": [\"%s\", \"%s\", \"%s\", \"%s\", \"%s\"], \"id\":4}",
-         rpc_user, work->job_id, xnonce2str, ntimestr, noncestr );
-   free( xnonce2str );
-}
-
-#define min(a,b) (a>b ? (b) :(a))
-
-// merkle root handled here, no need for gen_merkle_root gate target
-void blake2b_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
-{
-    uchar merkle_root[64] = { 0 };
-    uint32_t extraheader[32] = { 0 };
-    int headersize = 0;
-    size_t t;
-    int i;
-
-    // merkle root
-    memcpy( merkle_root, sctx->job.coinbase, 32 );
-    headersize = min( (int)sctx->job.coinbase_size - 32, sizeof(extraheader) );
-    memcpy( extraheader, &sctx->job.coinbase[32], headersize );
-    // Increment extranonce2 
-    for ( t = 0; t < sctx->xnonce2_size && !( ++sctx->job.xnonce2[t] ); t++ );
-    // Assemble block header 
-    memset( g_work->data, 0, sizeof(g_work->data) );
-//    g_work->data[0] = le32dec( sctx->job.version );
-//    for ( i = 0; i < 8; i++ )
-//       g_work->data[1 + i] = le32dec( (uint32_t *) sctx->job.prevhash + i );
-    for ( i = 0; i < 8; i++ )
-       g_work->data[i] = ((uint32_t*)sctx->job.prevhash)[7-i];
-//    for ( i = 0; i < 8; i++ )
-//       g_work->data[9 + i] = be32dec( (uint32_t *) merkle_root + i );
-    g_work->data[8]  = 0; // nonce
-    g_work->data[9]  = swab32( extraheader[0] ) | ( rand() & 0xf0 );
-    g_work->data[10] = be32dec( sctx->job.ntime );
-    g_work->data[11] = be32dec( sctx->job.nbits );
-    for ( i = 0; i < 8; i++ )
-       g_work->data[12+i] = ( (uint32_t*)merkle_root )[i];
-}
-
-#undef min
-
-void blake2b_get_new_work( struct work* work, struct work* g_work, int thr_id,
-                           uint32_t* end_nonce_ptr, bool clean_job )
-{
-   const int wkcmp_sz = 32;  // bytes
-   const int wkcmp_off = 32 + 16; 
-   uint32_t *nonceptr = algo_gate.get_nonceptr( work->data );
-
-   if ( memcmp( &work->data[ wkcmp_off ], &g_work->data[ wkcmp_off ], wkcmp_sz )
-      && ( clean_job || ( *nonceptr >= *end_nonce_ptr ) 
-      || strcmp( work->job_id, g_work->job_id ) ) )
-   {
-      work_free( work );
-      work_copy( work, g_work );
-      *nonceptr = ( 0xffffffffU / opt_n_threads ) * thr_id;
-      if ( opt_randomize )
-         *nonceptr += ( (rand() *4 ) & UINT32_MAX ) / opt_n_threads;
-      *end_nonce_ptr = ( 0xffffffffU / opt_n_threads ) * (thr_id+1) - 0x20;
-   }
-   else
-       ++(*nonceptr);
-
-   // suprnova job_id check without data/target/height change...
-   // we just may have copied new g_wwork to work so why this test here?
-//   if (  have_stratum && strcmp( work->job_id, g_work->job_id ) )
-      // exit thread loop
-//      continue;
-//   else
-//   {
-//      nonceptr[1] += 0x10;
-//      nonceptr[1] |= thr_id;
-//   }
-}
-
-bool blake2b_ready_to_mine( struct work* work, struct stratum_ctx* stratum,
-                           int thr_id )
-{
-   if ( have_stratum && strcmp( stratum->job.job_id, work->job_id ) )
-      // need to regen g_work..
-      return false;
-   // extradata: prevent duplicates
-   work->data[ 8     ] += 0x10;
-   work->data[ 8 + 1 ] |= thr_id;
-   return true;
-}
-
-double blake2b_get_max64() { return 0x1fffffLL; }
-
-bool register_blake2b_algo( algo_gate_t* gate )
-{
-  algo_not_tested();
-  gate->ntime_index   = 10;
-  gate->nbits_index   = 11;
-  gate->nonce_index   =  8;
-  gate->work_cmp_size = 32;
-  gate->scanhash              = (void*)&scanhash_blake2b;
-  gate->hash                  = (void*)&blake2b_hash;
-  gate->calc_network_diff     = (void*)&blake2b_calc_network_diff;
-  gate->build_stratum_request = (void*)&blake2b_be_build_stratum_request;
-  gate->work_decode           = (void*)&std_be_work_decode;
-  gate->submit_getwork_result = (void*)&std_be_submit_getwork_result;
-  gate->build_extraheader     = (void*)&blake2b_build_extraheader;
-  gate->get_new_work          = (void*)&blake2b_get_new_work;
-  gate->get_max64             = (void*)&blake2b_get_max64;
-  gate->ready_to_mine         = (void*)&blake2b_ready_to_mine;
-  have_gbt = false;
-  return true;
-}

algo/blake/blake2s-gate.c

@@ -1,12 +1,5 @@
 #include "blake2s-gate.h"
 
-
-// changed to get_max64_0x3fffffLL in cpuminer-multi-decred
-int64_t blake2s_get_max64 ()
-{
-   return 0x7ffffLL;
-}
-
 bool register_blake2s_algo( algo_gate_t* gate )
 {
 #if defined(BLAKE2S_8WAY)
@@ -19,8 +12,7 @@ bool register_blake2s_algo( algo_gate_t* gate )
   gate->scanhash  = (void*)&scanhash_blake2s;
   gate->hash      = (void*)&blake2s_hash;
 #endif
-  gate->get_max64 = (void*)&blake2s_get_max64;
-  gate->optimizations = SSE42_OPT | AVX2_OPT;
+  gate->optimizations = SSE2_OPT | AVX2_OPT;
   return true;
 };
 

algo/blake/blake2s-gate.h

@@ -4,7 +4,8 @@
 #include <stdint.h>
 #include "algo-gate-api.h"
 
-#if defined(__SSE4_2__)
+//#if defined(__SSE4_2__)
+#if defined(__SSE2__)
   #define BLAKE2S_4WAY
 #endif
 #if defined(__AVX2__)

algo/blake/blake2s-hash-4way.c

@@ -17,13 +17,16 @@
 #include <string.h>
 #include <stdio.h>
 
-#if defined(__SSE4_2__)
+//#if defined(__SSE4_2__)
+#if defined(__SSE2__)
 
+/*
 static const uint32_t blake2s_IV[8] =
 {
 	0x6A09E667UL, 0xBB67AE85UL, 0x3C6EF372UL, 0xA54FF53AUL,
 	0x510E527FUL, 0x9B05688CUL, 0x1F83D9ABUL, 0x5BE0CD19UL
 };
+*/
 
 static const uint8_t blake2s_sigma[10][16] =
 {
@@ -39,6 +42,7 @@ static const uint8_t blake2s_sigma[10][16] =
 	{ 10,  2,  8,  4,  7,  6,  1,  5, 15, 11,  9, 14,  3, 12, 13 , 0 } ,
 };
 
+
 // define a constant for initial param.
 
 int blake2s_4way_init( blake2s_4way_state *S, const uint8_t outlen )
@@ -57,8 +61,18 @@ int blake2s_4way_init( blake2s_4way_state *S, const uint8_t outlen )
    memset( P->personal, 0, sizeof( P->personal ) );
 
    memset( S, 0, sizeof( blake2s_4way_state ) );
-   for( int i = 0; i < 8; ++i )
-      S->h[i] = _mm_set1_epi32( blake2s_IV[i] );
+
+   S->h[0] = m128_const1_64( 0x6A09E6676A09E667ULL );
+   S->h[1] = m128_const1_64( 0xBB67AE85BB67AE85ULL );
+   S->h[2] = m128_const1_64( 0x3C6EF3723C6EF372ULL );
+   S->h[3] = m128_const1_64( 0xA54FF53AA54FF53AULL );
+   S->h[4] = m128_const1_64( 0x510E527F510E527FULL );
+   S->h[5] = m128_const1_64( 0x9B05688C9B05688CULL );
+   S->h[6] = m128_const1_64( 0x1F83D9AB1F83D9ABULL );
+   S->h[7] = m128_const1_64( 0x5BE0CD195BE0CD19ULL );
+   
+//   for( int i = 0; i < 8; ++i )
+//      S->h[i] = _mm_set1_epi32( blake2s_IV[i] );
 
    uint32_t *p = ( uint32_t * )( P );
 
@@ -76,41 +90,45 @@ int blake2s_4way_compress( blake2s_4way_state *S, const __m128i* block )
    memcpy_128( m, block, 16 );
    memcpy_128( v, S->h, 8 );
 
-   v[ 8] = _mm_set1_epi32( blake2s_IV[0] );
-   v[ 9] = _mm_set1_epi32( blake2s_IV[1] );
-   v[10] = _mm_set1_epi32( blake2s_IV[2] );
-   v[11] = _mm_set1_epi32( blake2s_IV[3] );
+   v[ 8] = m128_const1_64( 0x6A09E6676A09E667ULL );
+   v[ 9] = m128_const1_64( 0xBB67AE85BB67AE85ULL );
+   v[10] = m128_const1_64( 0x3C6EF3723C6EF372ULL );
+   v[11] = m128_const1_64( 0xA54FF53AA54FF53AULL );
    v[12] = _mm_xor_si128( _mm_set1_epi32( S->t[0] ),
-                          _mm_set1_epi32( blake2s_IV[4] ) );
+                          m128_const1_64( 0x510E527F510E527FULL ) );
    v[13] = _mm_xor_si128( _mm_set1_epi32( S->t[1] ),
-                          _mm_set1_epi32( blake2s_IV[5] ) );
+                          m128_const1_64( 0x9B05688C9B05688CULL ) );
    v[14] = _mm_xor_si128( _mm_set1_epi32( S->f[0] ),
-                          _mm_set1_epi32( blake2s_IV[6] ) );
+                          m128_const1_64( 0x1F83D9AB1F83D9ABULL ) );
    v[15] = _mm_xor_si128( _mm_set1_epi32( S->f[1] ),
-                          _mm_set1_epi32( blake2s_IV[7] ) );
+                          m128_const1_64( 0x5BE0CD195BE0CD19ULL ) );
 
-#define G4W(r,i,a,b,c,d) \
+#define G4W( sigma0, sigma1, a, b, c, d ) \
 do { \
-   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ blake2s_sigma[r][2*i+0] ] ); \
+   uint8_t s0 = sigma0; \
+   uint8_t s1 = sigma1; \
+   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ s0 ] ); \
    d = mm128_ror_32( _mm_xor_si128( d, a ), 16 ); \
    c = _mm_add_epi32( c, d ); \
    b = mm128_ror_32( _mm_xor_si128( b, c ), 12 ); \
-   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ blake2s_sigma[r][2*i+1] ] ); \
+   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ s1 ] ); \
    d = mm128_ror_32( _mm_xor_si128( d, a ),  8 ); \
    c = _mm_add_epi32( c, d ); \
    b = mm128_ror_32( _mm_xor_si128( b, c ),  7 ); \
 } while(0)
 
+
 #define ROUND4W(r)  \
 do { \
-   G4W( r, 0, v[ 0], v[ 4], v[ 8], v[12] ); \
-   G4W( r, 1, v[ 1], v[ 5], v[ 9], v[13] ); \
-   G4W( r, 2, v[ 2], v[ 6], v[10], v[14] ); \
-   G4W( r, 3, v[ 3], v[ 7], v[11], v[15] ); \
-   G4W( r, 4, v[ 0], v[ 5], v[10], v[15] ); \
-   G4W( r, 5, v[ 1], v[ 6], v[11], v[12] ); \
-   G4W( r, 6, v[ 2], v[ 7], v[ 8], v[13] ); \
-   G4W( r, 7, v[ 3], v[ 4], v[ 9], v[14] ); \
+   uint8_t *sigma = (uint8_t*)&blake2s_sigma[r]; \
+   G4W( sigma[ 0], sigma[ 1], v[ 0], v[ 4], v[ 8], v[12] ); \
+   G4W( sigma[ 2], sigma[ 3], v[ 1], v[ 5], v[ 9], v[13] ); \
+   G4W( sigma[ 4], sigma[ 5], v[ 2], v[ 6], v[10], v[14] ); \
+   G4W( sigma[ 6], sigma[ 7], v[ 3], v[ 7], v[11], v[15] ); \
+   G4W( sigma[ 8], sigma[ 9], v[ 0], v[ 5], v[10], v[15] ); \
+   G4W( sigma[10], sigma[11], v[ 1], v[ 6], v[11], v[12] ); \
+   G4W( sigma[12], sigma[13], v[ 2], v[ 7], v[ 8], v[13] ); \
+   G4W( sigma[14], sigma[15], v[ 3], v[ 4], v[ 9], v[14] ); \
 } while(0)
 
    ROUND4W( 0 );
@@ -132,26 +150,47 @@ do { \
    return 0;
 }
 
+// There is a problem that can't be resolved internally.
+// If the last block is a full 64 bytes it should not be compressed in
+// update but left for final. However, when streaming, it isn't known
+// which block is last. There may be a subsequent call to update to add
+// more data.
+//
+// The reference code handled this by juggling 2 blocks at a time at
+// a significant performance penalty.
+//
+// Instead a new function is introduced called full_blocks which combines
+// update and final and is to be used in non-streaming mode where the data
+// is a multiple of 64 bytes.
+// 
+// Supported:
+//    64 + 16 bytes  (blake2s with midstate optimization)
+//    80 bytes without midstate (blake2s without midstate optimization)
+//    Any multiple of 64 bytes in one shot (x25x)
+//
+// Unsupported:
+//    Stream of 64 byte blocks one at a time.   
+//
+// use for part blocks or when streaming more data
 int blake2s_4way_update( blake2s_4way_state *S, const void *in,
                          uint64_t inlen )
 {
-  __m128i *input = (__m128i*)in;
-  __m128i *buf = (__m128i*)S->buf;
-  const int bsize = BLAKE2S_BLOCKBYTES;
+   __m128i *input = (__m128i*)in;
+   __m128i *buf = (__m128i*)S->buf;
 
    while( inlen > 0 )
    {
       size_t left = S->buflen;
-      if( inlen >= bsize - left )
+      if( inlen >= BLAKE2S_BLOCKBYTES - left )
       {
-         memcpy_128( buf + (left>>2), input, (bsize - left) >> 2 );
-         S->buflen += bsize - left;
+         memcpy_128( buf + (left>>2), input, (BLAKE2S_BLOCKBYTES - left) >> 2 );
+         S->buflen += BLAKE2S_BLOCKBYTES - left;
          S->t[0] += BLAKE2S_BLOCKBYTES;
          S->t[1] += ( S->t[0] < BLAKE2S_BLOCKBYTES );
          blake2s_4way_compress( S, buf ); 
          S->buflen = 0;
-         input += ( bsize >> 2 );
-         inlen -= bsize;
+         input += ( BLAKE2S_BLOCKBYTES >> 2 );
+         inlen -= BLAKE2S_BLOCKBYTES;
       }
       else
       {
@@ -183,8 +222,45 @@ int blake2s_4way_final( blake2s_4way_state *S, void *out, uint8_t outlen )
    return 0;
 }
 
+// Update and final when inlen is a multiple of 64 bytes
+int blake2s_4way_full_blocks( blake2s_4way_state *S, void *out,
+                              const void *input, uint64_t inlen )
+{
+    __m128i *in = (__m128i*)input;
+    __m128i *buf = (__m128i*)S->buf;
+
+    while( inlen > BLAKE2S_BLOCKBYTES )
+    {
+       memcpy_128( buf, in, BLAKE2S_BLOCKBYTES >> 2 );
+       S->buflen = BLAKE2S_BLOCKBYTES;
+       inlen -= BLAKE2S_BLOCKBYTES;
+       S->t[0] += BLAKE2S_BLOCKBYTES;
+       S->t[1] += ( S->t[0] < BLAKE2S_BLOCKBYTES );
+       blake2s_4way_compress( S, buf );
+       S->buflen = 0;
+       in += ( BLAKE2S_BLOCKBYTES >> 2 );
+    }
+
+    // last block
+    memcpy_128( buf, in, BLAKE2S_BLOCKBYTES >> 2 );
+    S->buflen = BLAKE2S_BLOCKBYTES;
+    S->t[0] += S->buflen;
+    S->t[1] += ( S->t[0] < S->buflen );
+    if ( S->last_node )  S->f[1] = ~0U;
+    S->f[0] = ~0U;
+    blake2s_4way_compress( S, buf );
+
+    for ( int i = 0; i < 8; ++i )
+      casti_m128i( out, i ) = S->h[ i ];
+    return 0;
+}
+
 #if defined(__AVX2__)
 
+// The commented code below is slower on Intel but faster on
+// Zen1 AVX2. It's also faster than Zen1 AVX.
+// Ryzen gen2 is unknown at this time.
+ 
 int blake2s_8way_compress( blake2s_8way_state *S, const __m256i *block )
 {
    __m256i m[16];
@@ -193,6 +269,23 @@ int blake2s_8way_compress( blake2s_8way_state *S, const __m256i *block )
    memcpy_256( m, block, 16 );
    memcpy_256( v, S->h, 8 );
 
+   v[ 8] = m256_const1_64( 0x6A09E6676A09E667ULL );
+   v[ 9] = m256_const1_64( 0xBB67AE85BB67AE85ULL );
+   v[10] = m256_const1_64( 0x3C6EF3723C6EF372ULL );
+   v[11] = m256_const1_64( 0xA54FF53AA54FF53AULL );
+   v[12] = _mm256_xor_si256( _mm256_set1_epi32( S->t[0] ),
+                          m256_const1_64( 0x510E527F510E527FULL ) );
+
+   v[13] = _mm256_xor_si256( _mm256_set1_epi32( S->t[1] ),
+                          m256_const1_64( 0x9B05688C9B05688CULL ) );
+
+   v[14] = _mm256_xor_si256( _mm256_set1_epi32( S->f[0] ),
+                          m256_const1_64( 0x1F83D9AB1F83D9ABULL ) );
+
+   v[15] = _mm256_xor_si256( _mm256_set1_epi32( S->f[1] ),
+                          m256_const1_64( 0x5BE0CD195BE0CD19ULL ) );
+
+/*
    v[ 8] = _mm256_set1_epi32( blake2s_IV[0] );
    v[ 9] = _mm256_set1_epi32( blake2s_IV[1] );
    v[10] = _mm256_set1_epi32( blake2s_IV[2] );
@@ -206,6 +299,7 @@ int blake2s_8way_compress( blake2s_8way_state *S, const __m256i *block )
    v[15] = _mm256_xor_si256( _mm256_set1_epi32( S->f[1] ),
                              _mm256_set1_epi32( blake2s_IV[7] ) );
 
+
 #define G8W(r,i,a,b,c,d) \
 do { \
    a = _mm256_add_epi32( _mm256_add_epi32( a, b ), \
@@ -219,7 +313,36 @@ do { \
    c = _mm256_add_epi32( c, d ); \
    b = mm256_ror_32( _mm256_xor_si256( b, c ),  7 ); \
 } while(0)
+*/
 
+#define G8W( sigma0, sigma1, a, b, c, d) \
+do { \
+   uint8_t s0 = sigma0; \
+   uint8_t s1 = sigma1; \
+   a = _mm256_add_epi32( _mm256_add_epi32( a, b ), m[ s0 ] ); \
+   d = mm256_ror_32( _mm256_xor_si256( d, a ), 16 ); \
+   c = _mm256_add_epi32( c, d ); \
+   b = mm256_ror_32( _mm256_xor_si256( b, c ), 12 ); \
+   a = _mm256_add_epi32( _mm256_add_epi32( a, b ), m[ s1 ] ); \
+   d = mm256_ror_32( _mm256_xor_si256( d, a ),  8 ); \
+   c = _mm256_add_epi32( c, d ); \
+   b = mm256_ror_32( _mm256_xor_si256( b, c ),  7 ); \
+} while(0)
+
+#define ROUND8W(r)  \
+do { \
+   uint8_t *sigma = (uint8_t*)&blake2s_sigma[r]; \
+   G8W( sigma[ 0], sigma[ 1], v[ 0], v[ 4], v[ 8], v[12] ); \
+   G8W( sigma[ 2], sigma[ 3], v[ 1], v[ 5], v[ 9], v[13] ); \
+   G8W( sigma[ 4], sigma[ 5], v[ 2], v[ 6], v[10], v[14] ); \
+   G8W( sigma[ 6], sigma[ 7], v[ 3], v[ 7], v[11], v[15] ); \
+   G8W( sigma[ 8], sigma[ 9], v[ 0], v[ 5], v[10], v[15] ); \
+   G8W( sigma[10], sigma[11], v[ 1], v[ 6], v[11], v[12] ); \
+   G8W( sigma[12], sigma[13], v[ 2], v[ 7], v[ 8], v[13] ); \
+   G8W( sigma[14], sigma[15], v[ 3], v[ 4], v[ 9], v[14] ); \
+} while(0)
+
+/*
 #define ROUND8W(r)  \
 do { \
    G8W( r, 0, v[ 0], v[ 4], v[ 8], v[12] ); \
@@ -231,6 +354,7 @@ do { \
    G8W( r, 6, v[ 2], v[ 7], v[ 8], v[13] ); \
    G8W( r, 7, v[ 3], v[ 4], v[ 9], v[14] ); \
 } while(0)
+*/
 
    ROUND8W( 0 );
    ROUND8W( 1 );
@@ -267,8 +391,18 @@ int blake2s_8way_init( blake2s_8way_state *S, const uint8_t outlen )
    memset( P->personal, 0, sizeof( P->personal ) );
 
    memset( S, 0, sizeof( blake2s_8way_state ) );
-   for( int i = 0; i < 8; ++i )
-      S->h[i] = _mm256_set1_epi32( blake2s_IV[i] );
+   S->h[0] = m256_const1_64( 0x6A09E6676A09E667ULL );
+   S->h[1] = m256_const1_64( 0xBB67AE85BB67AE85ULL );
+   S->h[2] = m256_const1_64( 0x3C6EF3723C6EF372ULL );
+   S->h[3] = m256_const1_64( 0xA54FF53AA54FF53AULL );
+   S->h[4] = m256_const1_64( 0x510E527F510E527FULL );
+   S->h[5] = m256_const1_64( 0x9B05688C9B05688CULL );
+   S->h[6] = m256_const1_64( 0x1F83D9AB1F83D9ABULL );
+   S->h[7] = m256_const1_64( 0x5BE0CD195BE0CD19ULL );
+
+
+//   for( int i = 0; i < 8; ++i )
+//      S->h[i] = _mm256_set1_epi32( blake2s_IV[i] );
 
    uint32_t *p = ( uint32_t * )( P );
 

algo/blake/blake2s-hash-4way.h

@@ -14,7 +14,8 @@
 #ifndef __BLAKE2S_HASH_4WAY_H__
 #define __BLAKE2S_HASH_4WAY_H__ 1
 
-#if defined(__SSE4_2__)
+//#if defined(__SSE4_2__)
+#if defined(__SSE2__)
 
 #include "simd-utils.h"
 
@@ -63,7 +64,7 @@ typedef struct __blake2s_nway_param
 ALIGN( 64 ) typedef struct __blake2s_4way_state
 {
    __m128i h[8];
-   uint8_t  buf[ BLAKE2S_BLOCKBYTES * 4 ];
+   uint8_t  buf[ 2 * BLAKE2S_BLOCKBYTES * 4 ];
    uint32_t t[2];
    uint32_t f[2];
    size_t   buflen;
@@ -80,7 +81,7 @@ int blake2s_4way_final( blake2s_4way_state *S, void *out, uint8_t outlen );
 ALIGN( 64 ) typedef struct __blake2s_8way_state
 {
    __m256i h[8];
-   uint8_t  buf[ BLAKE2S_BLOCKBYTES * 8 ];
+   uint8_t  buf[ 2 * BLAKE2S_BLOCKBYTES * 8 ];
    uint32_t t[2];
    uint32_t f[2];
    size_t   buflen;
@@ -91,6 +92,9 @@ int blake2s_8way_init( blake2s_8way_state *S, const uint8_t outlen );
 int blake2s_8way_update( blake2s_8way_state *S, const void *in,
                          uint64_t inlen );
 int blake2s_8way_final( blake2s_8way_state *S, void *out, uint8_t outlen );
+int blake2s_4way_full_blocks( blake2s_4way_state *S, void *out,
+                              const void *input, uint64_t inlen );
+
 
 #endif
 

algo/blake/blake2s.c

@@ -70,18 +70,3 @@ int scanhash_blake2s( struct work *work,
 
 	return 0;
 }
-/*
-// changed to get_max64_0x3fffffLL in cpuminer-multi-decred
-int64_t blake2s_get_max64 ()
-{
-   return 0x7ffffLL;
-}
-
-bool register_blake2s_algo( algo_gate_t* gate )
-{
-  gate->scanhash  = (void*)&scanhash_blake2s;
-  gate->hash      = (void*)&blake2s_hash;
-  gate->get_max64 = (void*)&blake2s_get_max64;
-  return true;
-};
-*/

algo/blake/blake512-hash-4way.c

@@ -307,12 +307,12 @@ static const sph_u64 CB[16] = {
 
 #define GB_4WAY(m0, m1, c0, c1, a, b, c, d)   do { \
    a = _mm256_add_epi64( _mm256_add_epi64( _mm256_xor_si256( \
-                 _mm256_set_epi64x( c1, c1, c1, c1 ), m0 ), b ), a ); \
+                 _mm256_set1_epi64x( c1 ), m0 ), b ), a ); \
    d = mm256_ror_64( _mm256_xor_si256( d, a ), 32 ); \
    c = _mm256_add_epi64( c, d ); \
    b = mm256_ror_64( _mm256_xor_si256( b, c ), 25 ); \
    a = _mm256_add_epi64( _mm256_add_epi64( _mm256_xor_si256( \
-                 _mm256_set_epi64x( c0, c0, c0, c0 ), m1 ), b ), a ); \
+                 _mm256_set1_epi64x( c0 ), m1 ), b ), a ); \
    d = mm256_ror_64( _mm256_xor_si256( d, a ), 16 ); \
    c = _mm256_add_epi64( c, d ); \
    b = mm256_ror_64( _mm256_xor_si256( b, c ), 11 ); \
@@ -403,7 +403,9 @@ static const sph_u64 CB[16] = {
 	__m256i M[16]; \
 	__m256i V0, V1, V2, V3, V4, V5, V6, V7; \
 	__m256i V8, V9, VA, VB, VC, VD, VE, VF; \
-	unsigned r; \
+   const __m256i shuff_bswap64 = m256_const2_64( 0x08090a0b0c0d0e0f, \
+                                                 0x0001020304050607 ) \
+   unsigned r; \
 	V0 = H0; \
 	V1 = H1; \
 	V2 = H2; \
@@ -412,53 +414,53 @@ static const sph_u64 CB[16] = {
 	V5 = H5; \
 	V6 = H6; \
 	V7 = H7; \
-   V8 = _mm256_xor_si256( S0, _mm256_set_epi64x( CB0, CB0, CB0, CB0 ) ); \
-   V9 = _mm256_xor_si256( S1, _mm256_set_epi64x( CB1, CB1, CB1, CB1 ) ); \
-   VA = _mm256_xor_si256( S2, _mm256_set_epi64x( CB2, CB2, CB2, CB2 ) ); \
-   VB = _mm256_xor_si256( S3, _mm256_set_epi64x( CB3, CB3, CB3, CB3 ) ); \
-   VC = _mm256_xor_si256( _mm256_set_epi64x( T0, T0, T0, T0 ), \
-                          _mm256_set_epi64x( CB4, CB4, CB4, CB4 ) ); \
-   VD = _mm256_xor_si256( _mm256_set_epi64x( T0, T0, T0, T0 ), \
-                          _mm256_set_epi64x( CB5, CB5, CB5, CB5 ) ); \
-   VE = _mm256_xor_si256( _mm256_set_epi64x( T1, T1, T1, T1 ), \
-                          _mm256_set_epi64x( CB6, CB6, CB6, CB6 ) ); \
-   VF = _mm256_xor_si256( _mm256_set_epi64x( T1, T1, T1, T1 ), \
-                          _mm256_set_epi64x( CB7, CB7, CB7, CB7 ) ); \
-	M[0x0] = mm256_bswap_64( *(buf+0) ); \
-	M[0x1] = mm256_bswap_64( *(buf+1) ); \
-	M[0x2] = mm256_bswap_64( *(buf+2) ); \
-	M[0x3] = mm256_bswap_64( *(buf+3) ); \
-	M[0x4] = mm256_bswap_64( *(buf+4) ); \
-	M[0x5] = mm256_bswap_64( *(buf+5) ); \
-	M[0x6] = mm256_bswap_64( *(buf+6) ); \
-	M[0x7] = mm256_bswap_64( *(buf+7) ); \
-	M[0x8] = mm256_bswap_64( *(buf+8) ); \
-	M[0x9] = mm256_bswap_64( *(buf+9) ); \
-	M[0xA] = mm256_bswap_64( *(buf+10) ); \
-	M[0xB] = mm256_bswap_64( *(buf+11) ); \
-	M[0xC] = mm256_bswap_64( *(buf+12) ); \
-	M[0xD] = mm256_bswap_64( *(buf+13) ); \
-	M[0xE] = mm256_bswap_64( *(buf+14) ); \
-	M[0xF] = mm256_bswap_64( *(buf+15) ); \
+   V8 = _mm256_xor_si256( S0, _mm256_set1_epi64x( CB0 ) ); \
+   V9 = _mm256_xor_si256( S1, _mm256_set1_epi64x( CB1 ) ); \
+   VA = _mm256_xor_si256( S2, _mm256_set1_epi64x( CB2 ) ); \
+   VB = _mm256_xor_si256( S3, _mm256_set1_epi64x( CB3 ) ); \
+   VC = _mm256_xor_si256( _mm256_set1_epi64x( T0 ), \
+                          _mm256_set1_epi64x( CB4 ) ); \
+   VD = _mm256_xor_si256( _mm256_set1_epi64x( T0 ), \
+                          _mm256_set1_epi64x( CB5 ) ); \
+   VE = _mm256_xor_si256( _mm256_set1_epi64x( T1 ), \
+                          _mm256_set1_epi64x( CB6 ) ); \
+   VF = _mm256_xor_si256( _mm256_set1_epi64x( T1 ), \
+                          _mm256_set1_epi64x( CB7, CB7, CB7, CB7 ) ); \
+   M[0x0] = _mm256_shuffle_epi8( *(buf+ 0), shuff_bswap64 ); \
+	M[0x1] = _mm256_shuffle_epi8( *(buf+ 1), shuff_bswap64 ); \
+	M[0x2] = _mm256_shuffle_epi8( *(buf+ 2), shuff_bswap64 ); \
+	M[0x3] = _mm256_shuffle_epi8( *(buf+ 3), shuff_bswap64 ); \
+	M[0x4] = _mm256_shuffle_epi8( *(buf+ 4), shuff_bswap64 ); \
+	M[0x5] = _mm256_shuffle_epi8( *(buf+ 5), shuff_bswap64 ); \
+	M[0x6] = _mm256_shuffle_epi8( *(buf+ 6), shuff_bswap64 ); \
+	M[0x7] = _mm256_shuffle_epi8( *(buf+ 7), shuff_bswap64 ); \
+	M[0x8] = _mm256_shuffle_epi8( *(buf+ 8), shuff_bswap64 ); \
+	M[0x9] = _mm256_shuffle_epi8( *(buf+ 9), shuff_bswap64 ); \
+	M[0xA] = _mm256_shuffle_epi8( *(buf+10), shuff_bswap64 ); \
+	M[0xB] = _mm256_shuffle_epi8( *(buf+11), shuff_bswap64 ); \
+	M[0xC] = _mm256_shuffle_epi8( *(buf+12), shuff_bswap64 ); \
+	M[0xD] = _mm256_shuffle_epi8( *(buf+13), shuff_bswap64 ); \
+	M[0xE] = _mm256_shuffle_epi8( *(buf+14), shuff_bswap64 ); \
+	M[0xF] = _mm256_shuffle_epi8( *(buf+15), shuff_bswap64 ); \
 	for (r = 0; r < 16; r ++) \
 		ROUND_B_4WAY(r); \
-        H0 = _mm256_xor_si256( _mm256_xor_si256( \
+   H0 = _mm256_xor_si256( _mm256_xor_si256( \
                     _mm256_xor_si256( S0, V0 ), V8 ), H0 ); \
-        H1 = _mm256_xor_si256( _mm256_xor_si256( \
+   H1 = _mm256_xor_si256( _mm256_xor_si256( \
                     _mm256_xor_si256( S1, V1 ), V9 ), H1 ); \
-        H2 = _mm256_xor_si256( _mm256_xor_si256( \
+   H2 = _mm256_xor_si256( _mm256_xor_si256( \
                     _mm256_xor_si256( S2, V2 ), VA ), H2 ); \
-        H3 = _mm256_xor_si256( _mm256_xor_si256( \
+   H3 = _mm256_xor_si256( _mm256_xor_si256( \
                     _mm256_xor_si256( S3, V3 ), VB ), H3 ); \
-        H4 = _mm256_xor_si256( _mm256_xor_si256( \
+   H4 = _mm256_xor_si256( _mm256_xor_si256( \
                     _mm256_xor_si256( S0, V4 ), VC ), H4 ); \
-        H5 = _mm256_xor_si256( _mm256_xor_si256( \
+   H5 = _mm256_xor_si256( _mm256_xor_si256( \
                     _mm256_xor_si256( S1, V5 ), VD ), H5 ); \
-        H6 = _mm256_xor_si256( _mm256_xor_si256( \
+   H6 = _mm256_xor_si256( _mm256_xor_si256( \
                     _mm256_xor_si256( S2, V6 ), VE ), H6 ); \
-        H7 = _mm256_xor_si256( _mm256_xor_si256( \
+   H7 = _mm256_xor_si256( _mm256_xor_si256( \
                     _mm256_xor_si256( S3, V7 ), VF ), H7 ); \
-	} while (0)
+} while (0)
 
 #else
 
@@ -479,20 +481,19 @@ static const sph_u64 CB[16] = {
   V5 = H5; \
   V6 = H6; \
   V7 = H7; \
-  V8 = _mm256_xor_si256( S0, _mm256_set1_epi64x( CB0 ) );  \
-  V9 = _mm256_xor_si256( S1, _mm256_set1_epi64x( CB1 ) );  \
-  VA = _mm256_xor_si256( S2, _mm256_set1_epi64x( CB2 ) );  \
-  VB = _mm256_xor_si256( S3, _mm256_set1_epi64x( CB3 ) );  \
+  V8 = _mm256_xor_si256( S0, m256_const1_64( CB0 ) );  \
+  V9 = _mm256_xor_si256( S1, m256_const1_64( CB1 ) );  \
+  VA = _mm256_xor_si256( S2, m256_const1_64( CB2 ) );  \
+  VB = _mm256_xor_si256( S3, m256_const1_64( CB3 ) );  \
   VC = _mm256_xor_si256( _mm256_set1_epi64x( T0 ), \
-                         _mm256_set1_epi64x( CB4 ) );  \
+                         m256_const1_64( CB4 ) );  \
   VD = _mm256_xor_si256( _mm256_set1_epi64x( T0 ), \
-                         _mm256_set1_epi64x( CB5 ) );  \
+                         m256_const1_64( CB5 ) );  \
   VE = _mm256_xor_si256( _mm256_set1_epi64x( T1 ), \
-                         _mm256_set1_epi64x( CB6 ) );  \
+                         m256_const1_64( CB6 ) );  \
   VF = _mm256_xor_si256( _mm256_set1_epi64x( T1 ), \
-                         _mm256_set1_epi64x( CB7 ) );  \
-  shuf_bswap64 = _mm256_set_epi64x( 0x08090a0b0c0d0e0f, 0x0001020304050607, \
-                                    0x08090a0b0c0d0e0f, 0x0001020304050607 ); \
+                         m256_const1_64( CB7 ) );  \
+  shuf_bswap64 = m256_const2_64( 0x08090a0b0c0d0e0f, 0x0001020304050607 ); \
   M0 = _mm256_shuffle_epi8( *(buf+ 0), shuf_bswap64 ); \
   M1 = _mm256_shuffle_epi8( *(buf+ 1), shuf_bswap64 ); \
   M2 = _mm256_shuffle_epi8( *(buf+ 2), shuf_bswap64 ); \
@@ -544,14 +545,14 @@ blake64_4way_init( blake_4way_big_context *sc, const sph_u64 *iv,
               const sph_u64 *salt )
 {
    __m256i zero = m256_zero;
-   casti_m256i( sc->H, 0 ) = _mm256_set1_epi64x( iv[0] );
-   casti_m256i( sc->H, 1 ) = _mm256_set1_epi64x( iv[1] );
-   casti_m256i( sc->H, 2 ) = _mm256_set1_epi64x( iv[2] );
-   casti_m256i( sc->H, 3 ) = _mm256_set1_epi64x( iv[3] );
-   casti_m256i( sc->H, 4 ) = _mm256_set1_epi64x( iv[4] );
-   casti_m256i( sc->H, 5 ) = _mm256_set1_epi64x( iv[5] );
-   casti_m256i( sc->H, 6 ) = _mm256_set1_epi64x( iv[6] );
-   casti_m256i( sc->H, 7 ) = _mm256_set1_epi64x( iv[7] );
+   casti_m256i( sc->H, 0 ) = m256_const1_64( 0x6A09E667F3BCC908 );
+   casti_m256i( sc->H, 1 ) = m256_const1_64( 0xBB67AE8584CAA73B );
+   casti_m256i( sc->H, 2 ) = m256_const1_64( 0x3C6EF372FE94F82B );
+   casti_m256i( sc->H, 3 ) = m256_const1_64( 0xA54FF53A5F1D36F1 );
+   casti_m256i( sc->H, 4 ) = m256_const1_64( 0x510E527FADE682D1 );
+   casti_m256i( sc->H, 5 ) = m256_const1_64( 0x9B05688C2B3E6C1F );
+   casti_m256i( sc->H, 6 ) = m256_const1_64( 0x1F83D9ABFB41BD6B );
+   casti_m256i( sc->H, 7 ) = m256_const1_64( 0x5BE0CD19137E2179 );
 
    casti_m256i( sc->S, 0 ) = zero;
    casti_m256i( sc->S, 1 ) = zero;
@@ -620,7 +621,7 @@ blake64_4way_close( blake_4way_big_context *sc,
    bit_len = ((unsigned)ptr << 3);
    z = 0x80 >> n;
    zz = ((ub & -z) | z) & 0xFF;
-   buf[ptr>>3] = _mm256_set_epi64x( zz, zz, zz, zz );
+   buf[ptr>>3] = _mm256_set1_epi64x( zz );
    tl = sc->T0 + bit_len;
    th = sc->T1;
    if (ptr == 0 )
@@ -642,11 +643,9 @@ blake64_4way_close( blake_4way_big_context *sc,
        memset_zero_256( buf + (ptr>>3) + 1, (104-ptr) >> 3 );
        if ( out_size_w64 == 8 )
           buf[(104>>3)] = _mm256_or_si256( buf[(104>>3)],
-                                 _mm256_set1_epi64x( 0x0100000000000000ULL ) );
-       *(buf+(112>>3)) = mm256_bswap_64(
-                                    _mm256_set_epi64x( th, th, th, th ) );
-       *(buf+(120>>3)) = mm256_bswap_64(
-                                    _mm256_set_epi64x( tl, tl, tl, tl ) );
+                                 m256_const1_64( 0x0100000000000000ULL ) );
+       *(buf+(112>>3)) = _mm256_set1_epi64x( bswap_64( th ) );
+       *(buf+(120>>3)) = _mm256_set1_epi64x( bswap_64( tl ) );
 
        blake64_4way( sc, buf + (ptr>>3), 128 - ptr );
    }
@@ -659,11 +658,9 @@ blake64_4way_close( blake_4way_big_context *sc,
        sc->T1 = SPH_C64(0xFFFFFFFFFFFFFFFFULL);
        memset_zero_256( buf, 112>>3 ); 
        if ( out_size_w64 == 8 )
-           buf[104>>3] = _mm256_set1_epi64x( 0x0100000000000000ULL );
-       *(buf+(112>>3)) = mm256_bswap_64(
-                                    _mm256_set_epi64x( th, th, th, th ) );
-       *(buf+(120>>3)) = mm256_bswap_64(
-                                    _mm256_set_epi64x( tl, tl, tl, tl ) );
+           buf[104>>3] = m256_const1_64( 0x0100000000000000ULL );
+       *(buf+(112>>3)) = _mm256_set1_epi64x( bswap_64( th ) );
+       *(buf+(120>>3)) = _mm256_set1_epi64x( bswap_64( tl ) );
 
        blake64_4way( sc, buf, 128 );
    }

algo/blake/blakecoin-gate.c

@@ -1,13 +1,6 @@
 #include "blakecoin-gate.h"
 #include <memory.h>
 
-// changed to get_max64_0x3fffffLL in cpuminer-multi-decred
-int64_t blakecoin_get_max64 ()
-{
-  return 0x7ffffLL;
-//  return 0x3fffffLL;
-}
-
 // vanilla uses default gen merkle root, otherwise identical to blakecoin
 bool register_vanilla_algo( algo_gate_t* gate )
 {
@@ -23,7 +16,6 @@ bool register_vanilla_algo( algo_gate_t* gate )
   gate->hash     = (void*)&blakecoinhash;
 #endif
   gate->optimizations = SSE42_OPT | AVX2_OPT;
-  gate->get_max64 = (void*)&blakecoin_get_max64;
   return true;
 }
 

algo/blake/blakecoin.c

@@ -93,33 +93,3 @@ int scanhash_blakecoin( struct work *work, uint32_t max_nonce,
 	return 0;
 }
 
-/*
-void blakecoin_gen_merkle_root ( char* merkle_root, struct stratum_ctx* sctx )
-{
- SHA256( sctx->job.coinbase, (int)sctx->job.coinbase_size, merkle_root );
-}
-*/
-/*
-// changed to get_max64_0x3fffffLL in cpuminer-multi-decred
-int64_t blakecoin_get_max64 ()
-{
-  return 0x7ffffLL;
-}
-
-// vanilla uses default gen merkle root, otherwise identical to blakecoin
-bool register_vanilla_algo( algo_gate_t* gate )
-{
-    gate->scanhash = (void*)&scanhash_blakecoin;
-    gate->hash     = (void*)&blakecoinhash;
-    gate->get_max64 = (void*)&blakecoin_get_max64;
-    blakecoin_init( &blake_init_ctx );
-    return true;
-}
-
-bool register_blakecoin_algo( algo_gate_t* gate )
-{
-  register_vanilla_algo( gate );
-  gate->gen_merkle_root = (void*)&SHA256_gen_merkle_root;
-  return true;
-}
-*/

algo/blake/decred-gate.c

@@ -38,7 +38,7 @@ void decred_decode_extradata( struct work* work, uint64_t* net_blocks )
    if (!have_longpoll && work->height > *net_blocks + 1)
    {
       char netinfo[64] = { 0 };
-      if (opt_showdiff && net_diff > 0.)
+      if ( net_diff > 0. )
       {
          if (net_diff != work->targetdiff)
             sprintf(netinfo, ", diff %.3f, target %.1f", net_diff,
@@ -116,7 +116,7 @@ void decred_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
    // block header suffix from coinb2 (stake version)
    memcpy( &g_work->data[44],
            &sctx->job.coinbase[ sctx->job.coinbase_size-4 ], 4 );
-   sctx->bloc_height = g_work->data[32];
+   sctx->block_height = g_work->data[32];
    //applog_hex(work->data, 180);
    //applog_hex(&work->data[36], 36);
 }
@@ -154,7 +154,6 @@ bool register_decred_algo( algo_gate_t* gate )
 #endif
   gate->optimizations = AVX2_OPT;
   gate->get_nonceptr          = (void*)&decred_get_nonceptr;
-  gate->get_max64             = (void*)&get_max64_0x3fffffLL;
   gate->decode_extra_data     = (void*)&decred_decode_extradata;
   gate->build_stratum_request = (void*)&decred_be_build_stratum_request;
   gate->work_decode           = (void*)&std_be_work_decode;

algo/blake/decred.c

@@ -143,7 +143,7 @@ void decred_decode_extradata( struct work* work, uint64_t* net_blocks )
    if (!have_longpoll && work->height > *net_blocks + 1)
    {
       char netinfo[64] = { 0 };
-      if (opt_showdiff && net_diff > 0.)
+      if (net_diff > 0.)
       {
          if (net_diff != work->targetdiff)
 	    sprintf(netinfo, ", diff %.3f, target %.1f", net_diff,
@@ -269,7 +269,6 @@ bool register_decred_algo( algo_gate_t* gate )
   gate->scanhash              = (void*)&scanhash_decred;
   gate->hash                  = (void*)&decred_hash;
   gate->get_nonceptr          = (void*)&decred_get_nonceptr;
-  gate->get_max64             = (void*)&get_max64_0x3fffffLL;
   gate->decode_extra_data     = (void*)&decred_decode_extradata;
   gate->build_stratum_request = (void*)&decred_be_build_stratum_request;
   gate->work_decode           = (void*)&std_be_work_decode;

algo/blake/pentablake-gate.c

@@ -10,7 +10,6 @@ bool register_pentablake_algo( algo_gate_t* gate )
     gate->hash      = (void*)&pentablakehash;
 #endif
     gate->optimizations = AVX2_OPT;
-    gate->get_max64 = (void*)&get_max64_0x3ffff;
     return true;
 };
 

algo/blake/sph_blake2b.c

@@ -103,7 +103,6 @@ static void blake2b_compress( sph_blake2b_ctx *ctx, int last )
 	v[13] ^= ctx->t[1];                 // high 64 bits
 	if (last)                           // last block flag set ?
 		v[14] = ~v[14];
-
 	for (i = 0; i < 16; i++)            // get little-endian words
 		m[i] = B2B_GET64(&ctx->b[8 * i]);
 
@@ -184,7 +183,8 @@ void sph_blake2b_final( sph_blake2b_ctx *ctx, void *out )
 
 	while (ctx->c < 128)                // fill up with zeros
 		ctx->b[ctx->c++] = 0;
-	blake2b_compress(ctx, 1);           // final block flag = 1
+
+   blake2b_compress(ctx, 1);           // final block flag = 1
 
 	// little endian convert and store
 	for (i = 0; i < ctx->outlen; i++) {

algo/bmw/bmw-hash-4way.h

@@ -62,7 +62,7 @@ typedef struct {
 
 typedef bmw_4way_small_context bmw256_4way_context;
 
-void bmw256_4way_init(void *cc);
+void bmw256_4way_init( bmw256_4way_context *ctx );
 
 void bmw256_4way(void *cc, const void *data, size_t len);
 

algo/bmw/bmw256-hash-4way.c

@@ -48,7 +48,7 @@ extern "C"{
 #if defined(__SSE2__)
 
 // BMW-256 4 way 32
-
+/*
 static const uint32_t IV256[] = {
 	0x40414243, 0x44454647,
 	0x48494A4B, 0x4C4D4E4F,
@@ -59,6 +59,7 @@ static const uint32_t IV256[] = {
 	0x70717273, 0x74757677,
 	0x78797A7B, 0x7C7D7E7F
 };
+*/
 
 #define ss0(x) \
    _mm_xor_si128( _mm_xor_si128( _mm_srli_epi32( (x), 1), \
@@ -462,13 +463,30 @@ static const __m128i final_s[16] =
    { 0xaaaaaaafaaaaaaaf, 0xaaaaaaafaaaaaaaf }
 };
 */
-static void
-bmw32_4way_init(bmw_4way_small_context *sc, const sph_u32 *iv)
+void bmw256_4way_init( bmw256_4way_context *ctx )
 {
-   for ( int i = 0; i < 16; i++ )
-      sc->H[i] = _mm_set1_epi32( iv[i] );
-   sc->ptr = 0;
-   sc->bit_count = 0;
+   ctx->H[ 0] = m128_const1_64( 0x4041424340414243 );
+   ctx->H[ 1] = m128_const1_64( 0x4445464744454647 );
+   ctx->H[ 2] = m128_const1_64( 0x48494A4B48494A4B );
+   ctx->H[ 3] = m128_const1_64( 0x4C4D4E4F4C4D4E4F );
+   ctx->H[ 4] = m128_const1_64( 0x5051525350515253 );
+   ctx->H[ 5] = m128_const1_64( 0x5455565754555657 );
+   ctx->H[ 6] = m128_const1_64( 0x58595A5B58595A5B );
+   ctx->H[ 7] = m128_const1_64( 0x5C5D5E5F5C5D5E5F );
+   ctx->H[ 8] = m128_const1_64( 0x6061626360616263 );
+   ctx->H[ 9] = m128_const1_64( 0x6465666764656667 );
+   ctx->H[10] = m128_const1_64( 0x68696A6B68696A6B );
+   ctx->H[11] = m128_const1_64( 0x6C6D6E6F6C6D6E6F );
+   ctx->H[12] = m128_const1_64( 0x7071727370717273 );
+   ctx->H[13] = m128_const1_64( 0x7475767774757677 );
+   ctx->H[14] = m128_const1_64( 0x78797A7B78797A7B );
+   ctx->H[15] = m128_const1_64( 0x7C7D7E7F7C7D7E7F );
+
+
+//   for ( int i = 0; i < 16; i++ )
+//      sc->H[i] = _mm_set1_epi32( iv[i] );
+   ctx->ptr = 0;
+   ctx->bit_count = 0;
 }
 
 static void
@@ -525,7 +543,7 @@ bmw32_4way_close(bmw_4way_small_context *sc, unsigned ub, unsigned n,
 
    buf = sc->buf;
    ptr = sc->ptr;
-   buf[ ptr>>2 ] = _mm_set1_epi32( 0x80 );
+   buf[ ptr>>2 ] = m128_const1_64( 0x0000008000000080 );
    ptr += 4;
    h = sc->H;
 
@@ -551,11 +569,13 @@ bmw32_4way_close(bmw_4way_small_context *sc, unsigned ub, unsigned n,
       casti_m128i( dst, u ) = h1[v];
 }
 
+/*
 void
 bmw256_4way_init(void *cc)
 {
 	bmw32_4way_init(cc, IV256);
 }
+*/
 
 void
 bmw256_4way(void *cc, const void *data, size_t len)
@@ -1003,25 +1023,24 @@ static const __m256i final_s8[16] =
 
 void bmw256_8way_init( bmw256_8way_context *ctx )
 {
-   ctx->H[ 0] = _mm256_set1_epi32( IV256[ 0] );
-   ctx->H[ 1] = _mm256_set1_epi32( IV256[ 1] );
-   ctx->H[ 2] = _mm256_set1_epi32( IV256[ 2] );
-   ctx->H[ 3] = _mm256_set1_epi32( IV256[ 3] );
-   ctx->H[ 4] = _mm256_set1_epi32( IV256[ 4] );
-   ctx->H[ 5] = _mm256_set1_epi32( IV256[ 5] );
-   ctx->H[ 6] = _mm256_set1_epi32( IV256[ 6] );
-   ctx->H[ 7] = _mm256_set1_epi32( IV256[ 7] );
-   ctx->H[ 8] = _mm256_set1_epi32( IV256[ 8] );
-   ctx->H[ 9] = _mm256_set1_epi32( IV256[ 9] );
-   ctx->H[10] = _mm256_set1_epi32( IV256[10] );
-   ctx->H[11] = _mm256_set1_epi32( IV256[11] );
-   ctx->H[12] = _mm256_set1_epi32( IV256[12] );
-   ctx->H[13] = _mm256_set1_epi32( IV256[13] );
-   ctx->H[14] = _mm256_set1_epi32( IV256[14] );
-   ctx->H[15] = _mm256_set1_epi32( IV256[15] );
+   ctx->H[ 0] = m256_const1_64( 0x4041424340414243 );
+   ctx->H[ 1] = m256_const1_64( 0x4445464744454647 );
+   ctx->H[ 2] = m256_const1_64( 0x48494A4B48494A4B );
+   ctx->H[ 3] = m256_const1_64( 0x4C4D4E4F4C4D4E4F );
+   ctx->H[ 4] = m256_const1_64( 0x5051525350515253 );
+   ctx->H[ 5] = m256_const1_64( 0x5455565754555657 );
+   ctx->H[ 6] = m256_const1_64( 0x58595A5B58595A5B );
+   ctx->H[ 7] = m256_const1_64( 0x5C5D5E5F5C5D5E5F );
+   ctx->H[ 8] = m256_const1_64( 0x6061626360616263 );
+   ctx->H[ 9] = m256_const1_64( 0x6465666764656667 );
+   ctx->H[10] = m256_const1_64( 0x68696A6B68696A6B );
+   ctx->H[11] = m256_const1_64( 0x6C6D6E6F6C6D6E6F );
+   ctx->H[12] = m256_const1_64( 0x7071727370717273 );
+   ctx->H[13] = m256_const1_64( 0x7475767774757677 );
+   ctx->H[14] = m256_const1_64( 0x78797A7B78797A7B );
+   ctx->H[15] = m256_const1_64( 0x7C7D7E7F7C7D7E7F );
    ctx->ptr       = 0;
    ctx->bit_count = 0;
-
 }
 
 void bmw256_8way( bmw256_8way_context *ctx, const void *data, size_t len )
@@ -1074,7 +1093,7 @@ void bmw256_8way_close( bmw256_8way_context *ctx, void *dst )
 
    buf = ctx->buf;
    ptr = ctx->ptr;
-   buf[ ptr>>2 ] = _mm256_set1_epi32( 0x80 );
+   buf[ ptr>>2 ] = m256_const1_64( 0x0000008000000080 );
    ptr += 4;
    h = ctx->H;
 
@@ -1089,7 +1108,6 @@ void bmw256_8way_close( bmw256_8way_context *ctx, void *dst )
    buf[ (buf_size - 8) >> 2 ] = _mm256_set1_epi32( ctx->bit_count );
    buf[ (buf_size - 4) >> 2 ] = m256_zero;
 
-
    compress_small_8way( buf, h, h2 );
 
    for ( u = 0; u < 16; u ++ )

algo/bmw/bmw512-gate.c

@@ -1,12 +1,9 @@
 #include "bmw512-gate.h"
 
-int64_t bmw512_get_max64() { return 0x7ffffLL; }
-
 bool register_bmw512_algo( algo_gate_t* gate )
 {
   gate->optimizations = AVX2_OPT;
-  gate->set_target      = (void*)&alt_set_target;
-  gate->get_max64       = (void*)&bmw512_get_max64;
+  opt_target_factor = 256.0;
 #if defined (BMW512_4WAY)
   gate->scanhash  = (void*)&scanhash_bmw512_4way;
   gate->hash      = (void*)&bmw512hash_4way;

algo/bmw/bmw512-hash-4way.c

@@ -961,8 +961,22 @@ static const __m256i final_b[16] =
 static void
 bmw64_4way_init( bmw_4way_big_context *sc, const sph_u64 *iv )
 {
-   for ( int i = 0; i < 16; i++ )
-      sc->H[i] = _mm256_set1_epi64x( iv[i] );
+   sc->H[ 0] = m256_const1_64( 0x8081828384858687 );
+   sc->H[ 1] = m256_const1_64( 0x88898A8B8C8D8E8F );
+   sc->H[ 2] = m256_const1_64( 0x9091929394959697 );
+   sc->H[ 3] = m256_const1_64( 0x98999A9B9C9D9E9F );
+   sc->H[ 4] = m256_const1_64( 0xA0A1A2A3A4A5A6A7 );
+   sc->H[ 5] = m256_const1_64( 0xA8A9AAABACADAEAF );
+   sc->H[ 6] = m256_const1_64( 0xB0B1B2B3B4B5B6B7 );
+   sc->H[ 7] = m256_const1_64( 0xB8B9BABBBCBDBEBF );
+   sc->H[ 8] = m256_const1_64( 0xC0C1C2C3C4C5C6C7 );
+   sc->H[ 9] = m256_const1_64( 0xC8C9CACBCCCDCECF );
+   sc->H[10] = m256_const1_64( 0xD0D1D2D3D4D5D6D7 );
+   sc->H[11] = m256_const1_64( 0xD8D9DADBDCDDDEDF );
+   sc->H[12] = m256_const1_64( 0xE0E1E2E3E4E5E6E7 );
+   sc->H[13] = m256_const1_64( 0xE8E9EAEBECEDEEEF );
+   sc->H[14] = m256_const1_64( 0xF0F1F2F3F4F5F6F7 );
+   sc->H[15] = m256_const1_64( 0xF8F9FAFBFCFDFEFF );
    sc->ptr = 0;
    sc->bit_count = 0;
 }
@@ -1014,13 +1028,11 @@ bmw64_4way_close(bmw_4way_big_context *sc, unsigned ub, unsigned n,
    __m256i *buf;
    __m256i h1[16], h2[16], *h;
    size_t ptr, u, v;
-   unsigned z;
    const int buf_size = 128;  // bytes of one lane, compatible with len
 
    buf = sc->buf;
    ptr = sc->ptr;
-   z = 0x80 >> n;
-   buf[ ptr>>3 ] = _mm256_set1_epi64x( z );
+   buf[ ptr>>3 ] = m256_const1_64( 0x80 );
    ptr += 8;
    h = sc->H;
 

algo/cryptonight/cryptolight.c

@@ -363,7 +363,6 @@ bool register_cryptolight_algo( algo_gate_t* gate )
   gate->scanhash  = (void*)&scanhash_cryptolight;
   gate->hash      = (void*)&cryptolight_hash;
   gate->hash_suw  = (void*)&cryptolight_hash; 
-  gate->get_max64 = (void*)&get_max64_0x40LL;
   return true;
 };
 

algo/cryptonight/cryptonight-common.c

@@ -111,7 +111,6 @@ bool register_cryptonight_algo( algo_gate_t* gate )
   gate->scanhash         = (void*)&scanhash_cryptonight;
   gate->hash             = (void*)&cryptonight_hash;
   gate->hash_suw         = (void*)&cryptonight_hash_suw;  
-  gate->get_max64        = (void*)&get_max64_0x40LL;
   return true;
 };
 
@@ -123,7 +122,6 @@ bool register_cryptonightv7_algo( algo_gate_t* gate )
   gate->scanhash      = (void*)&scanhash_cryptonight;
   gate->hash          = (void*)&cryptonight_hash;
   gate->hash_suw      = (void*)&cryptonight_hash_suw;
-  gate->get_max64     = (void*)&get_max64_0x40LL;
   return true;
 };
 

algo/cubehash/cube-hash-2way.c

@@ -7,7 +7,7 @@
 
 // 2x128
 
-/*
+
 // The result of hashing 10 rounds of initial data which consists of params
 // zero padded.
 static const uint64_t IV256[] =
@@ -25,7 +25,7 @@ static const uint64_t IV512[] =
 0x148FE485FCD398D9, 0xB64445321B017BEF, 0x2FF5781C6A536159, 0x0DBADEA991FA7934,
 0xA5A70E75D65C8A2B, 0xBC796576B1C62456, 0xE7989AF11921C8F7, 0xD43E3B447795D246
 };
-*/
+
 
 static void transform_2way( cube_2way_context *sp )
 {
@@ -97,39 +97,30 @@ static void transform_2way( cube_2way_context *sp )
 int cube_2way_init( cube_2way_context *sp, int hashbitlen, int rounds,
                     int blockbytes )
 {
-    __m128i* h = (__m128i*)sp->h;
+    __m256i *h = (__m256i*)sp->h;
+    __m128i *iv = (__m128i*)( hashbitlen == 512 ? (__m128i*)IV512
+                                                : (__m128i*)IV256 );
     sp->hashlen   = hashbitlen/128;
     sp->blocksize = blockbytes/16;
     sp->rounds    = rounds;
     sp->pos       = 0;
 
-    if ( hashbitlen == 512 )
-    {
-
-       h[ 0] = m128_const_64( 0x4167D83E2D538B8B, 0x50F494D42AEA2A61 );
-       h[ 2] = m128_const_64( 0x50AC5695CC39968E, 0xC701CF8C3FEE2313 );
-       h[ 4] = m128_const_64( 0x825B453797CF0BEF, 0xA647A8B34D42C787 );
-       h[ 6] = m128_const_64( 0xA23911AED0E5CD33, 0xF22090C4EEF864D2 );
-       h[ 8] = m128_const_64( 0xB64445321B017BEF, 0x148FE485FCD398D9 );
-       h[10] = m128_const_64( 0x0DBADEA991FA7934, 0x2FF5781C6A536159 );
-       h[12] = m128_const_64( 0xBC796576B1C62456, 0xA5A70E75D65C8A2B );
-       h[14] = m128_const_64( 0xD43E3B447795D246, 0xE7989AF11921C8F7 );
-       h[1] = h[ 0];  h[ 3] = h[ 2]; h[ 5] = h[ 4]; h[ 7] = h[ 6];
-       h[9] = h[ 8];  h[11] = h[10]; h[13] = h[12]; h[15] = h[14];
-    }
-    else
-    {
-       h[ 0] = m128_const_64( 0x35481EAE63117E71, 0xCCD6F29FEA2BD4B4 );
-       h[ 2] = m128_const_64( 0xF4CC12BE7E624131, 0xE5D94E6322512D5B );
-       h[ 4] = m128_const_64( 0x3361DA8CD0720C35, 0x42AF2070C2D0B696 );
-       h[ 6] = m128_const_64( 0x40E5FBAB4680AC00, 0x8EF8AD8328CCECA4 );
-       h[ 8] = m128_const_64( 0xF0B266796C859D41, 0x6107FBD5D89041C3 );
-       h[10] = m128_const_64( 0x93CB628565C892FD, 0x5FA2560309392549 );
-       h[12] = m128_const_64( 0x85254725774ABFDD, 0x9E4B4E602AF2B5AE );
-       h[14] = m128_const_64( 0xD6032C0A9CDAF8AF, 0x4AB6AAD615815AEB );
-       h[1] = h[ 0];  h[ 3] = h[ 2]; h[ 5] = h[ 4]; h[ 7] = h[ 6];
-       h[9] = h[ 8];  h[11] = h[10]; h[13] = h[12]; h[15] = h[14];
-    }
+    h[ 0] = m256_const1_128( iv[0] );
+    h[ 1] = m256_const1_128( iv[1] );
+    h[ 2] = m256_const1_128( iv[2] );
+    h[ 3] = m256_const1_128( iv[3] );
+    h[ 4] = m256_const1_128( iv[4] );
+    h[ 5] = m256_const1_128( iv[5] );
+    h[ 6] = m256_const1_128( iv[6] );
+    h[ 7] = m256_const1_128( iv[7] );
+    h[ 0] = m256_const1_128( iv[0] );
+    h[ 1] = m256_const1_128( iv[1] );
+    h[ 2] = m256_const1_128( iv[2] );
+    h[ 3] = m256_const1_128( iv[3] );
+    h[ 4] = m256_const1_128( iv[4] );
+    h[ 5] = m256_const1_128( iv[5] );
+    h[ 6] = m256_const1_128( iv[6] );
+    h[ 7] = m256_const1_128( iv[7] );
     
     return 0;
 }
@@ -164,11 +155,11 @@ int cube_2way_close( cube_2way_context *sp, void *output )
 
     // pos is zero for 64 byte data, 1 for 80 byte data.
     sp->h[ sp->pos ] = _mm256_xor_si256( sp->h[ sp->pos ],
-                                _mm256_set_epi32( 0,0,0,0x80,  0,0,0,0x80 ) );
+                                   m256_const2_64( 0, 0x0000000000000080 ) );
     transform_2way( sp );
 
     sp->h[7] = _mm256_xor_si256( sp->h[7],
-		                 _mm256_set_epi32( 1,0,0,0,  1,0,0,0 ) );
+                                   m256_const2_64( 0x0000000100000000, 0 ) );
 
     for ( i = 0; i < 10; ++i )           transform_2way( sp );
 
@@ -197,13 +188,13 @@ int cube_2way_update_close( cube_2way_context *sp, void *output,
 
     // pos is zero for 64 byte data, 1 for 80 byte data.
     sp->h[ sp->pos ] = _mm256_xor_si256( sp->h[ sp->pos ],
-                    _mm256_set_epi32( 0,0,0,0x80,  0,0,0,0x80 ) );
+                                    m256_const2_64( 0, 0x0000000000000080 ) );
     transform_2way( sp );
 
-    sp->h[7] = _mm256_xor_si256( sp->h[7], _mm256_set_epi32( 1,0,0,0,
-                                                             1,0,0,0 ) );
+    sp->h[7] = _mm256_xor_si256( sp->h[7],
+                                    m256_const2_64( 0x0000000100000000, 0 ) );
 
-    for ( i = 0; i < 10; ++i )            transform_2way( sp );
+    for ( i = 0; i < 10; ++i )    transform_2way( sp );
 
     memcpy( hash, sp->h, sp->hashlen<<5 );
     return 0;

algo/groestl/groestl.c

@@ -94,19 +94,13 @@ int scanhash_groestl( struct work *work, uint32_t max_nonce,
 	return 0;
 }
 
-void groestl_set_target( struct work* work, double job_diff )
-{
- work_set_target( work, job_diff / (256.0 * opt_diff_factor) );
-}
-
 bool register_dmd_gr_algo( algo_gate_t* gate )
 {
     init_groestl_ctx();
     gate->optimizations   = SSE2_OPT | AES_OPT;
     gate->scanhash        = (void*)&scanhash_groestl;
     gate->hash            = (void*)&groestlhash;
-    gate->set_target      = (void*)&groestl_set_target;
-    gate->get_max64       = (void*)&get_max64_0x3ffff;
+    opt_target_factor = 256.0;
     return true;
 };
 

algo/groestl/myr-groestl.c

@@ -10,7 +10,7 @@
 #else
   #include "aes_ni/hash-groestl.h"
 #endif
-#include "algo/sha/sph_sha2.h"
+#include <openssl/sha.h>
 
 typedef struct {
 #ifdef NO_AES_NI
@@ -18,7 +18,7 @@ typedef struct {
 #else
     hashState_groestl       groestl;
 #endif
-    sph_sha256_context sha;
+    SHA256_CTX              sha;
 } myrgr_ctx_holder;
 
 myrgr_ctx_holder myrgr_ctx;
@@ -28,15 +28,15 @@ void init_myrgr_ctx()
 #ifdef NO_AES_NI
      sph_groestl512_init( &myrgr_ctx.groestl );
 #else
-     init_groestl (&myrgr_ctx.groestl, 64 );
+     init_groestl ( &myrgr_ctx.groestl, 64 );
 #endif
-     sph_sha256_init(&myrgr_ctx.sha);
+     SHA256_Init( &myrgr_ctx.sha );
 }
 
 void myriad_hash(void *output, const void *input)
 {
-        myrgr_ctx_holder ctx;
-        memcpy( &ctx, &myrgr_ctx, sizeof(myrgr_ctx) );
+   myrgr_ctx_holder ctx;
+   memcpy( &ctx, &myrgr_ctx, sizeof(myrgr_ctx) );
 
  	uint32_t _ALIGN(32) hash[16];
 
@@ -44,23 +44,22 @@ void myriad_hash(void *output, const void *input)
 	sph_groestl512(&ctx.groestl, input, 80);
 	sph_groestl512_close(&ctx.groestl, hash);
 #else
-        update_groestl( &ctx.groestl, (char*)input, 640 );
-        final_groestl( &ctx.groestl, (char*)hash);
+   update_groestl( &ctx.groestl, (char*)input, 640 );
+   final_groestl( &ctx.groestl, (char*)hash);
 #endif
 
-	sph_sha256(&ctx.sha, hash, 64);
-	sph_sha256_close(&ctx.sha, hash);
+   SHA256_Update( &ctx.sha, (unsigned char*)hash, 64 );
+   SHA256_Final( (unsigned char*)hash, &ctx.sha );
 
 	memcpy(output, hash, 32);
 }
 
-int scanhash_myriad( struct work *work,
-	uint32_t max_nonce, uint64_t *hashes_done, struct thr_info *mythr)
+int scanhash_myriad( struct work *work, uint32_t max_nonce,
+                     uint64_t *hashes_done, struct thr_info *mythr )
 {
-        uint32_t *pdata = work->data;
-        uint32_t *ptarget = work->target;
-
 	uint32_t _ALIGN(64) endiandata[20];
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
 	const uint32_t first_nonce = pdata[19];
 	uint32_t nonce = first_nonce;
    int thr_id = mythr->id;  // thr_id arg is deprecated
@@ -89,15 +88,3 @@ int scanhash_myriad( struct work *work,
 	*hashes_done = pdata[19] - first_nonce + 1;
 	return 0;
 }
-/*
-bool register_myriad_algo( algo_gate_t* gate )
-{
-    gate->optimizations = SSE2_OPT | AES_OPT;
-    init_myrgr_ctx();
-    gate->scanhash = (void*)&scanhash_myriad;
-    gate->hash     = (void*)&myriadhash;
-//    gate->hash_alt = (void*)&myriadhash;
-    gate->get_max64 = (void*)&get_max64_0x3ffff;
-    return true;
-};
-*/

algo/groestl/myrgr-4way.c

@@ -8,7 +8,7 @@
 #include <string.h>
 
 #include "aes_ni/hash-groestl.h"
-#include "algo/sha/sha2-hash-4way.h"
+#include "algo/sha/sha-hash-4way.h"
 
 typedef struct {
     hashState_groestl       groestl;

algo/groestl/myrgr-gate.c

@@ -12,7 +12,6 @@ bool register_myriad_algo( algo_gate_t* gate )
   gate->hash      = (void*)&myriad_hash;
 #endif
   gate->optimizations = AES_OPT | AVX2_OPT;
-  gate->get_max64 = (void*)&get_max64_0x3ffff;
   return true;
 };
 

algo/groestl/myrgr-gate.h

@@ -4,7 +4,7 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__AVX2__) && defined(__AES__)
+#if defined(__AVX2__) && defined(__AES__) && !defined(__SHA__)
   #define MYRGR_4WAY
 #endif
 

algo/hamsi/hamsi-hash-4way.c

@@ -32,8 +32,6 @@
 
 #include <stddef.h>
 #include <string.h>
-
-//#include "miner.h"
 #include "hamsi-hash-4way.h"
 
 #if defined(__AVX2__)
@@ -100,7 +98,7 @@ extern "C"{
 #endif
 
 //#include "hamsi-helper-4way.c"
-
+/*
 static const sph_u32 IV512[] = {
 	SPH_C32(0x73746565), SPH_C32(0x6c706172), SPH_C32(0x6b204172),
 	SPH_C32(0x656e6265), SPH_C32(0x72672031), SPH_C32(0x302c2062),
@@ -109,7 +107,7 @@ static const sph_u32 IV512[] = {
 	SPH_C32(0x65766572), SPH_C32(0x6c65652c), SPH_C32(0x2042656c),
 	SPH_C32(0x6769756d)
 };
-
+*/
 static const sph_u32 alpha_n[] = {
 	SPH_C32(0xff00f0f0), SPH_C32(0xccccaaaa), SPH_C32(0xf0f0cccc),
 	SPH_C32(0xff00aaaa), SPH_C32(0xccccaaaa), SPH_C32(0xf0f0ff00),
@@ -138,6 +136,7 @@ static const sph_u32 alpha_f[] = {
 	SPH_C32(0xcaf9f9c0), SPH_C32(0x0ff0639c)
 };
 
+
 // imported from hamsi helper
 
 /* Note: this table lists bits within each byte from least
@@ -529,49 +528,34 @@ static const sph_u32 T512[64][16] = {
 	  SPH_C32(0xe7e00a94) }
 };
 
+
 #define INPUT_BIG \
 do { \
-  const __m256i zero = _mm256_setzero_si256(); \
   __m256i db = *buf; \
-  const sph_u32 *tp = &T512[0][0]; \
-  m0 = zero; \
-  m1 = zero; \
-  m2 = zero; \
-  m3 = zero; \
-  m4 = zero; \
-  m5 = zero; \
-  m6 = zero; \
-  m7 = zero; \
+  const uint64_t *tp = (uint64_t*)&T512[0][0];  \
+  m0 = m1 = m2 = m3 = m4 = m5 = m6 = m7 = m256_zero; \
   for ( int u = 0; u < 64; u++ ) \
   { \
      __m256i dm = _mm256_and_si256( db, m256_one_64 ) ; \
      dm = mm256_negate_32( _mm256_or_si256( dm, \
                          _mm256_slli_epi64( dm, 32 ) ) ); \
      m0 = _mm256_xor_si256( m0, _mm256_and_si256( dm, \
-                  _mm256_set_epi32( tp[0x1], tp[0x0], tp[0x1], tp[0x0], \
-                                    tp[0x1], tp[0x0], tp[0x1], tp[0x0] ) ) ); \
+                                          m256_const1_64( tp[0] ) ) ); \
      m1 = _mm256_xor_si256( m1, _mm256_and_si256( dm, \
-                  _mm256_set_epi32( tp[0x3], tp[0x2], tp[0x3], tp[0x2], \
-                                    tp[0x3], tp[0x2], tp[0x3], tp[0x2] ) ) ); \
+                                          m256_const1_64( tp[1] ) ) ); \
      m2 = _mm256_xor_si256( m2, _mm256_and_si256( dm, \
-                  _mm256_set_epi32( tp[0x5], tp[0x4], tp[0x5], tp[0x4], \
-                                    tp[0x5], tp[0x4], tp[0x5], tp[0x4] ) ) ); \
+                                          m256_const1_64( tp[2] ) ) ); \
      m3 = _mm256_xor_si256( m3, _mm256_and_si256( dm, \
-                  _mm256_set_epi32( tp[0x7], tp[0x6], tp[0x7], tp[0x6], \
-                                    tp[0x7], tp[0x6], tp[0x7], tp[0x6] ) ) ); \
+                                          m256_const1_64( tp[3] ) ) ); \
      m4 = _mm256_xor_si256( m4, _mm256_and_si256( dm, \
-                  _mm256_set_epi32( tp[0x9], tp[0x8], tp[0x9], tp[0x8], \
-                                    tp[0x9], tp[0x8], tp[0x9], tp[0x8] ) ) ); \
+                                          m256_const1_64( tp[4] ) ) ); \
      m5 = _mm256_xor_si256( m5, _mm256_and_si256( dm, \
-                  _mm256_set_epi32( tp[0xB], tp[0xA], tp[0xB], tp[0xA], \
-                                    tp[0xB], tp[0xA], tp[0xB], tp[0xA] ) ) ); \
+                                          m256_const1_64( tp[5] ) ) ); \
      m6 = _mm256_xor_si256( m6, _mm256_and_si256( dm, \
-                  _mm256_set_epi32( tp[0xD], tp[0xC], tp[0xD], tp[0xC], \
-                                    tp[0xD], tp[0xC], tp[0xD], tp[0xC] ) ) ); \
+                                          m256_const1_64( tp[6] ) ) ); \
      m7 = _mm256_xor_si256( m7, _mm256_and_si256( dm, \
-                  _mm256_set_epi32( tp[0xF], tp[0xE], tp[0xF], tp[0xE], \
-                                    tp[0xF], tp[0xE], tp[0xF], tp[0xE] ) ) ); \
-     tp += 0x10; \
+                                          m256_const1_64( tp[7] ) ) ); \
+     tp += 8; \
      db = _mm256_srli_epi64( db, 1 ); \
   } \
 } while (0)
@@ -662,55 +646,39 @@ do { \
 
 #define ROUND_BIG(rc, alpha) \
 do { \
-  __m256i t0, t1, t2, t3; \
-  s0 = _mm256_xor_si256( s0, _mm256_set_epi32( \
-        alpha[0x01] ^ (rc), alpha[0x00], alpha[0x01] ^ (rc), alpha[0x00], \
-        alpha[0x01] ^ (rc), alpha[0x00], alpha[0x01] ^ (rc), alpha[0x00] ) ); \
-  s1 = _mm256_xor_si256( s1, _mm256_set_epi32( \
-                     alpha[0x03], alpha[0x02], alpha[0x03], alpha[0x02], \
-                     alpha[0x03], alpha[0x02], alpha[0x03], alpha[0x02] ) ); \
-  s2 = _mm256_xor_si256( s2, _mm256_set_epi32( \
-                     alpha[0x05], alpha[0x04], alpha[0x05], alpha[0x04], \
-                     alpha[0x05], alpha[0x04], alpha[0x05], alpha[0x04] ) ); \
-  s3 = _mm256_xor_si256( s3, _mm256_set_epi32( \
-                     alpha[0x07], alpha[0x06], alpha[0x07], alpha[0x06], \
-                     alpha[0x07], alpha[0x06], alpha[0x07], alpha[0x06] ) ); \
-  s4 = _mm256_xor_si256( s4, _mm256_set_epi32( \
-                     alpha[0x09], alpha[0x08], alpha[0x09], alpha[0x08], \
-                     alpha[0x09], alpha[0x08], alpha[0x09], alpha[0x08] ) ); \
-  s5 = _mm256_xor_si256( s5, _mm256_set_epi32( \
-                     alpha[0x0B], alpha[0x0A], alpha[0x0B], alpha[0x0A], \
-                     alpha[0x0B], alpha[0x0A], alpha[0x0B], alpha[0x0A] ) ); \
-  s6 = _mm256_xor_si256( s6, _mm256_set_epi32( \
-                     alpha[0x0D], alpha[0x0C], alpha[0x0D], alpha[0x0C], \
-                     alpha[0x0D], alpha[0x0C], alpha[0x0D], alpha[0x0C] ) ); \
-  s7 = _mm256_xor_si256( s7, _mm256_set_epi32( \
-                     alpha[0x0F], alpha[0x0E], alpha[0x0F], alpha[0x0E], \
-                     alpha[0x0F], alpha[0x0E], alpha[0x0F], alpha[0x0E] ) ); \
-  s8 = _mm256_xor_si256( s8, _mm256_set_epi32( \
-                     alpha[0x11], alpha[0x10], alpha[0x11], alpha[0x10], \
-                     alpha[0x11], alpha[0x10], alpha[0x11], alpha[0x10] ) ); \
-  s9 = _mm256_xor_si256( s9, _mm256_set_epi32( \
-                     alpha[0x13], alpha[0x12], alpha[0x13], alpha[0x12], \
-                     alpha[0x13], alpha[0x12], alpha[0x13], alpha[0x12] ) ); \
-  sA = _mm256_xor_si256( sA, _mm256_set_epi32( \
-                     alpha[0x15], alpha[0x14], alpha[0x15], alpha[0x14], \
-                     alpha[0x15], alpha[0x14], alpha[0x15], alpha[0x14] ) ); \
-  sB = _mm256_xor_si256( sB, _mm256_set_epi32( \
-                     alpha[0x17], alpha[0x16], alpha[0x17], alpha[0x16], \
-                     alpha[0x17], alpha[0x16], alpha[0x17], alpha[0x16] ) ); \
-  sC = _mm256_xor_si256( sC, _mm256_set_epi32( \
-                     alpha[0x19], alpha[0x18], alpha[0x19], alpha[0x18], \
-                     alpha[0x19], alpha[0x18], alpha[0x19], alpha[0x18] ) ); \
-  sD = _mm256_xor_si256( sD, _mm256_set_epi32( \
-                     alpha[0x1B], alpha[0x1A], alpha[0x1B], alpha[0x1A], \
-                     alpha[0x1B], alpha[0x1A], alpha[0x1B], alpha[0x1A] ) ); \
-  sE = _mm256_xor_si256( sE, _mm256_set_epi32( \
-                     alpha[0x1D], alpha[0x1C], alpha[0x1D], alpha[0x1C], \
-                     alpha[0x1D], alpha[0x1C], alpha[0x1D], alpha[0x1C] ) ); \
-  sF = _mm256_xor_si256( sF, _mm256_set_epi32( \
-                     alpha[0x1F], alpha[0x1E], alpha[0x1F], alpha[0x1E], \
-                     alpha[0x1F], alpha[0x1E], alpha[0x1F], alpha[0x1E] ) ); \
+   __m256i t0, t1, t2, t3; \
+   s0 = _mm256_xor_si256( s0, m256_const1_64( \
+        ( ( (uint64_t)( (rc) ^ alpha[1] ) << 32 ) ) | (uint64_t)alpha[0] ) ); \
+   s1 = _mm256_xor_si256( s1, m256_const1_64( \
+        ( (uint64_t)alpha[ 3] << 32 ) | (uint64_t)alpha[ 2] ) ); \
+   s2 = _mm256_xor_si256( s2, m256_const1_64( \
+        ( (uint64_t)alpha[ 5] << 32 ) | (uint64_t)alpha[ 4] ) ); \
+   s3 = _mm256_xor_si256( s3, m256_const1_64( \
+        ( (uint64_t)alpha[ 7] << 32 ) | (uint64_t)alpha[ 6] ) ); \
+   s4 = _mm256_xor_si256( s4, m256_const1_64( \
+        ( (uint64_t)alpha[ 9] << 32 ) | (uint64_t)alpha[ 8] ) ); \
+   s5 = _mm256_xor_si256( s5, m256_const1_64( \
+        ( (uint64_t)alpha[11] << 32 ) | (uint64_t)alpha[10] ) ); \
+   s6 = _mm256_xor_si256( s6, m256_const1_64( \
+        ( (uint64_t)alpha[13] << 32 ) | (uint64_t)alpha[12] ) ); \
+   s7 = _mm256_xor_si256( s7, m256_const1_64( \
+        ( (uint64_t)alpha[15] << 32 ) | (uint64_t)alpha[14] ) ); \
+   s8 = _mm256_xor_si256( s8, m256_const1_64( \
+        ( (uint64_t)alpha[17] << 32 ) | (uint64_t)alpha[16] ) ); \
+   s9 = _mm256_xor_si256( s9, m256_const1_64( \
+        ( (uint64_t)alpha[19] << 32 ) | (uint64_t)alpha[18] ) ); \
+   sA = _mm256_xor_si256( sA, m256_const1_64( \
+        ( (uint64_t)alpha[21] << 32 ) | (uint64_t)alpha[20] ) ); \
+   sB = _mm256_xor_si256( sB, m256_const1_64( \
+        ( (uint64_t)alpha[23] << 32 ) | (uint64_t)alpha[22] ) ); \
+   sC = _mm256_xor_si256( sC, m256_const1_64( \
+        ( (uint64_t)alpha[25] << 32 ) | (uint64_t)alpha[24] ) ); \
+   sD = _mm256_xor_si256( sD, m256_const1_64( \
+        ( (uint64_t)alpha[27] << 32 ) | (uint64_t)alpha[26] ) ); \
+   sE = _mm256_xor_si256( sE, m256_const1_64( \
+        ( (uint64_t)alpha[29] << 32 ) | (uint64_t)alpha[28] ) ); \
+   sF = _mm256_xor_si256( sF, m256_const1_64( \
+        ( (uint64_t)alpha[31] << 32 ) | (uint64_t)alpha[30] ) ); \
 \
   SBOX( s0, s4, s8, sC ); \
   SBOX( s1, s5, s9, sD ); \
@@ -864,47 +832,22 @@ void hamsi_big_final( hamsi_4way_big_context *sc, __m256i *buf )
 void hamsi512_4way_init( hamsi_4way_big_context *sc )
 {
    sc->partial_len = 0;
-   sph_u32 lo, hi;
    sc->count_high = sc->count_low = 0;
-   for ( int i = 0; i < 8; i++ )
-   {
-      lo = 2*i;
-      hi = 2*i + 1;
-      sc->h[i] = _mm256_set_epi32( IV512[hi], IV512[lo], IV512[hi], IV512[lo],
-                                   IV512[hi], IV512[lo], IV512[hi], IV512[lo] );
-   }
+
+   sc->h[0] = m256_const1_64( 0x6c70617273746565 );
+   sc->h[1] = m256_const1_64( 0x656e62656b204172 );
+   sc->h[2] = m256_const1_64( 0x302c206272672031 );
+   sc->h[3] = m256_const1_64( 0x3434362c75732032 );
+   sc->h[4] = m256_const1_64( 0x3030312020422d33 );
+   sc->h[5] = m256_const1_64( 0x656e2d484c657576 );
+   sc->h[6] = m256_const1_64( 0x6c65652c65766572 );
+   sc->h[7] = m256_const1_64( 0x6769756d2042656c );
 }
 
 void hamsi512_4way( hamsi_4way_big_context *sc, const void *data, size_t len )
 {
    __m256i *vdata = (__m256i*)data;
 
-// It looks like the only way to get in here is if core was previously called
-// with a very small len
-// That's not likely even with 80 byte input so deprecate partial len
-/*
-   if ( sc->partial_len != 0 )
-   {
-      size_t mlen;
-
-      mlen = 8 - sc->partial_len;
-      if ( len < mlen )
-      {
-         memcpy_256( sc->partial + (sc->partial_len >> 3), data, len>>3 );
-         sc->partial_len += len;
-         return;
-      }
-      else
-      {
-         memcpy_256( sc->partial + (sc->partial_len >> 3), data, mlen>>3 );
-         len -= mlen;
-         vdata += mlen>>3;
-         hamsi_big( sc, sc->partial, 1 );
-         sc->partial_len = 0;
-      }
-   }
-*/
-
    hamsi_big( sc, vdata, len>>3 );
    vdata += ( (len& ~(size_t)7) >> 3 );
    len &= (size_t)7;
@@ -920,8 +863,9 @@ void hamsi512_4way_close( hamsi_4way_big_context *sc, void *dst )
    sph_enc32be( &ch, sc->count_high );
    sph_enc32be( &cl, sc->count_low + ( sc->partial_len << 3 ) );
    pad[0] =  _mm256_set_epi32( cl, ch, cl, ch, cl, ch, cl, ch );
-   sc->buf[0] = _mm256_set_epi32( 0UL, 0x80UL, 0UL, 0x80UL,
-                                  0UL, 0x80UL, 0UL, 0x80UL );
+   sc->buf[0] = m256_const1_64( 0x80 );
+//      sc->buf[0] = _mm256_set_epi32( 0UL, 0x80UL, 0UL, 0x80UL,
+//                                  0UL, 0x80UL, 0UL, 0x80UL );
    hamsi_big( sc, sc->buf, 1 );
    hamsi_big_final( sc, pad );
 

algo/hodl/hodl-gate.c

@@ -15,11 +15,6 @@ pthread_barrier_t hodl_barrier;
 // need to be passed.
 unsigned char *hodl_scratchbuf = NULL;
 
-void hodl_set_target( struct work* work, double diff )
-{
-     diff_to_target(work->target, diff / 8388608.0 );
-}
-
 void hodl_le_build_stratum_request( char* req, struct work* work,
                                     struct stratum_ctx *sctx ) 
 {
@@ -170,7 +165,6 @@ bool register_hodl_algo( algo_gate_t* gate )
   gate->scanhash              = (void*)&hodl_scanhash;
   gate->get_new_work          = (void*)&hodl_get_new_work;
   gate->longpoll_rpc_call     = (void*)&hodl_longpoll_rpc_call;
-  gate->set_target            = (void*)&hodl_set_target;
   gate->build_stratum_request = (void*)&hodl_le_build_stratum_request;
   gate->malloc_txs_request    = (void*)&hodl_malloc_txs_request;
   gate->build_block_header    = (void*)&hodl_build_block_header;
@@ -179,6 +173,7 @@ bool register_hodl_algo( algo_gate_t* gate )
   gate->work_cmp_size         = 76;
   hodl_scratchbuf = (unsigned char*)malloc( 1 << 30 );
   allow_getwork = false;
+  opt_target_factor = 8388608.0;
   return ( hodl_scratchbuf != NULL );
 }
 

algo/jh/jh-hash-4way.c

@@ -94,7 +94,7 @@ extern "C"{
 
 #define Sb(x0, x1, x2, x3, c) \
 do { \
-   __m256i cc = _mm256_set_epi64x( c, c, c, c ); \
+   __m256i cc = _mm256_set1_epi64x( c ); \
     x3 = mm256_not( x3 ); \
     x0 = _mm256_xor_si256( x0, _mm256_andnot_si256( x2, cc ) ); \
     tmp = _mm256_xor_si256( cc, _mm256_and_si256( x0, x1 ) ); \
@@ -246,18 +246,12 @@ do { \
 	} while (0)
 */
 
-#define W0(x)   Wz(x, _mm256_set_epi64x( 0x5555555555555555, \
-       0x5555555555555555, 0x5555555555555555, 0x5555555555555555 ), 1 )
-#define W1(x)   Wz(x, _mm256_set_epi64x( 0x3333333333333333, \
-       0x3333333333333333, 0x3333333333333333, 0x3333333333333333 ), 2 )
-#define W2(x)   Wz(x, _mm256_set_epi64x( 0x0F0F0F0F0F0F0F0F, \
-       0x0F0F0F0F0F0F0F0F, 0x0F0F0F0F0F0F0F0F, 0x0F0F0F0F0F0F0F0F ), 4 )
-#define W3(x)   Wz(x, _mm256_set_epi64x( 0x00FF00FF00FF00FF, \
-       0x00FF00FF00FF00FF, 0x00FF00FF00FF00FF, 0x00FF00FF00FF00FF ), 8 ) 
-#define W4(x)   Wz(x, _mm256_set_epi64x( 0x0000FFFF0000FFFF, \
-       0x0000FFFF0000FFFF, 0x0000FFFF0000FFFF, 0x0000FFFF0000FFFF ), 16 )
-#define W5(x)   Wz(x, _mm256_set_epi64x( 0x00000000FFFFFFFF, \
-       0x00000000FFFFFFFF, 0x00000000FFFFFFFF, 0x00000000FFFFFFFF ), 32 )
+#define W0(x)   Wz(x, m256_const1_64( 0x5555555555555555 ),  1 )
+#define W1(x)   Wz(x, m256_const1_64( 0x3333333333333333 ),  2 )
+#define W2(x)   Wz(x, m256_const1_64( 0x0F0F0F0F0F0F0F0F ),  4 )
+#define W3(x)   Wz(x, m256_const1_64( 0x00FF00FF00FF00FF ),  8 ) 
+#define W4(x)   Wz(x, m256_const1_64( 0x0000FFFF0000FFFF ), 16 )
+#define W5(x)   Wz(x, m256_const1_64( 0x00000000FFFFFFFF ), 32 )
 #define W6(x) \
 do { \
    __m256i t = x ## h; \
@@ -331,14 +325,14 @@ do { \
 	__m256i m2l = buf[5]; \
 	__m256i m3h = buf[6]; \
 	__m256i m3l = buf[7]; \
-        h0h = _mm256_xor_si256( h0h, m0h ); \
-        h0l = _mm256_xor_si256( h0l, m0l ); \
-        h1h = _mm256_xor_si256( h1h, m1h ); \
-        h1l = _mm256_xor_si256( h1l, m1l ); \
-        h2h = _mm256_xor_si256( h2h, m2h ); \
-        h2l = _mm256_xor_si256( h2l, m2l ); \
-        h3h = _mm256_xor_si256( h3h, m3h ); \
-        h3l = _mm256_xor_si256( h3l, m3l ); \
+   h0h = _mm256_xor_si256( h0h, m0h ); \
+   h0l = _mm256_xor_si256( h0l, m0l ); \
+   h1h = _mm256_xor_si256( h1h, m1h ); \
+   h1l = _mm256_xor_si256( h1l, m1l ); \
+   h2h = _mm256_xor_si256( h2h, m2h ); \
+   h2l = _mm256_xor_si256( h2l, m2l ); \
+   h3h = _mm256_xor_si256( h3h, m3h ); \
+   h3l = _mm256_xor_si256( h3l, m3l ); \
 
 #define INPUT_BUF2 \
    h4h = _mm256_xor_si256( h4h, m0h ); \
@@ -477,13 +471,48 @@ static const sph_u64 IV512[] = {
 
 #endif
 
-static void
-jh_4way_init( jh_4way_context *sc, const void *iv )
+void jh256_4way_init( jh_4way_context *sc )
 {
-    uint64_t *v = (uint64_t*)iv;
-    
-    for ( int i = 0; i < 16; i++ )
-        sc->H[i] = _mm256_set_epi64x( v[i], v[i], v[i], v[i] );
+    // bswapped IV256
+    sc->H[ 0] = m256_const1_64( 0xebd3202c41a398eb );
+    sc->H[ 1] = m256_const1_64( 0xc145b29c7bbecd92 );
+    sc->H[ 2] = m256_const1_64( 0xfac7d4609151931c );
+    sc->H[ 3] = m256_const1_64( 0x038a507ed6820026 );
+    sc->H[ 4] = m256_const1_64( 0x45b92677269e23a4 );
+    sc->H[ 5] = m256_const1_64( 0x77941ad4481afbe0 );
+    sc->H[ 6] = m256_const1_64( 0x7a176b0226abb5cd );
+    sc->H[ 7] = m256_const1_64( 0xa82fff0f4224f056 );
+    sc->H[ 8] = m256_const1_64( 0x754d2e7f8996a371 );
+    sc->H[ 9] = m256_const1_64( 0x62e27df70849141d );
+    sc->H[10] = m256_const1_64( 0x948f2476f7957627 );
+    sc->H[11] = m256_const1_64( 0x6c29804757b6d587 );
+    sc->H[12] = m256_const1_64( 0x6c0d8eac2d275e5c );
+    sc->H[13] = m256_const1_64( 0x0f7a0557c6508451 );
+    sc->H[14] = m256_const1_64( 0xea12247067d3e47b );
+    sc->H[15] = m256_const1_64( 0x69d71cd313abe389 );
+    sc->ptr = 0;
+    sc->block_count = 0;
+}
+
+void jh512_4way_init( jh_4way_context *sc )
+{
+    // bswapped IV512
+    sc->H[ 0] = m256_const1_64( 0x17aa003e964bd16f );
+    sc->H[ 1] = m256_const1_64( 0x43d5157a052e6a63 );
+    sc->H[ 2] = m256_const1_64( 0x0bef970c8d5e228a );
+    sc->H[ 3] = m256_const1_64( 0x61c3b3f2591234e9 );
+    sc->H[ 4] = m256_const1_64( 0x1e806f53c1a01d89 );
+    sc->H[ 5] = m256_const1_64( 0x806d2bea6b05a92a );
+    sc->H[ 6] = m256_const1_64( 0xa6ba7520dbcc8e58 );
+    sc->H[ 7] = m256_const1_64( 0xf73bf8ba763a0fa9 );
+    sc->H[ 8] = m256_const1_64( 0x694ae34105e66901 );
+    sc->H[ 9] = m256_const1_64( 0x5ae66f2e8e8ab546 );
+    sc->H[10] = m256_const1_64( 0x243c84c1d0a74710 );
+    sc->H[11] = m256_const1_64( 0x99c15a2db1716e3b );
+    sc->H[12] = m256_const1_64( 0x56f8b19decf657cf );
+    sc->H[13] = m256_const1_64( 0x56b116577c8806a7 );
+    sc->H[14] = m256_const1_64( 0xfb1785e6dffcc2e3 );
+    sc->H[15] = m256_const1_64( 0x4bdd8ccc78465a54 );
     sc->ptr = 0;
     sc->block_count = 0;
 }
@@ -542,7 +571,7 @@ jh_4way_close( jh_4way_context *sc, unsigned ub, unsigned n, void *dst,
    size_t numz, u;
    sph_u64 l0, l1, l0e, l1e;
 
-   buf[0] = _mm256_set_epi64x( 0x80, 0x80, 0x80, 0x80 );
+   buf[0] = m256_const1_64( 0x80ULL );
 
    if ( sc->ptr == 0 )
        numz = 48;
@@ -555,8 +584,8 @@ jh_4way_close( jh_4way_context *sc, unsigned ub, unsigned n, void *dst,
    l1 = SPH_T64(sc->block_count >> 55);
    sph_enc64be( &l0e, l0 );
    sph_enc64be( &l1e, l1 );
-   *(buf + (numz>>3)    ) = _mm256_set_epi64x( l1e, l1e, l1e, l1e );
-   *(buf + (numz>>3) + 1) = _mm256_set_epi64x( l0e, l0e, l0e, l0e ); 
+   *(buf + (numz>>3)    ) = _mm256_set1_epi64x( l1e );
+   *(buf + (numz>>3) + 1) = _mm256_set1_epi64x( l0e ); 
 
    jh_4way_core( sc, buf, numz + 16 );
 
@@ -566,11 +595,13 @@ jh_4way_close( jh_4way_context *sc, unsigned ub, unsigned n, void *dst,
     memcpy_256( dst256, buf, 8 );
 }
 
+/*
 void
 jh256_4way_init(void *cc)
 {
-	jh_4way_init(cc, IV256);
+	jhs_4way_init(cc, IV256);
 }
+*/
 
 void
 jh256_4way(void *cc, const void *data, size_t len)
@@ -584,11 +615,13 @@ jh256_4way_close(void *cc, void *dst)
 	jh_4way_close(cc, 0, 0, dst, 8, IV256);
 }
 
+/*
 void
 jh512_4way_init(void *cc)
 {
-	jh_4way_init(cc, IV512);
+	jhb_4way_init(cc, IV512);
 }
+*/
 
 void
 jh512_4way(void *cc, const void *data, size_t len)

algo/jh/jh-hash-4way.h

@@ -79,13 +79,13 @@ typedef jh_4way_context jh256_4way_context;
 
 typedef jh_4way_context jh512_4way_context;
 
-void jh256_4way_init(void *cc);
+void jh256_4way_init( jh_4way_context *sc);
 
 void jh256_4way(void *cc, const void *data, size_t len);
 
 void jh256_4way_close(void *cc, void *dst);
 
-void jh512_4way_init(void *cc);
+void jh512_4way_init( jh_4way_context *sc );
 
 void jh512_4way(void *cc, const void *data, size_t len);
 

algo/jh/jha-gate.c

@@ -12,7 +12,7 @@ bool register_jha_algo( algo_gate_t* gate )
   gate->hash             = (void*)&jha_hash;
 #endif
   gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT;
-  gate->set_target       = (void*)&scrypt_set_target;
+  opt_target_factor = 65536.0;
   return true;
 };
 

algo/keccak/keccak-4way.c

@@ -39,10 +39,10 @@ int scanhash_keccak_4way( struct work *work, uint32_t max_nonce,
       keccakhash_4way( hash, vdata );
 
       for ( int lane = 0; lane < 4; lane++ )
-      if ( ( ( hash7[ lane<<1 ] & 0xFFFFFF00 ) == 0 ) )
+      if ( ( hash7[ lane<<1 ] & 0xFFFFFF00 ) == 0 )
       {
           extr_lane_4x64( lane_hash, hash, lane, 256 );
-          if ( fulltest( lane_hash, ptarget ) )
+          if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
           {
               pdata[19] = n + lane;
               submit_lane_solution( work, lane_hash, mythr, lane );

algo/keccak/keccak-gate.c

@@ -1,18 +1,11 @@
 #include "keccak-gate.h"
 
-void keccak_set_target( struct work* work, double job_diff )
-{
-  work_set_target( work, job_diff / (128.0 * opt_diff_factor) );
-}
-
-int64_t keccak_get_max64() { return 0x7ffffLL; }
 
 bool register_keccak_algo( algo_gate_t* gate )
 {
   gate->optimizations = AVX2_OPT;
   gate->gen_merkle_root = (void*)&SHA256_gen_merkle_root;
-  gate->set_target      = (void*)&keccak_set_target;
-  gate->get_max64       = (void*)&keccak_get_max64;
+  opt_target_factor = 128.0;
 #if defined (KECCAK_4WAY)
   gate->scanhash  = (void*)&scanhash_keccak_4way;
   gate->hash      = (void*)&keccakhash_4way;
@@ -23,17 +16,11 @@ bool register_keccak_algo( algo_gate_t* gate )
   return true;
 };
 
-void keccakc_set_target( struct work* work, double job_diff )
-{
-  work_set_target( work, job_diff / (256.0 * opt_diff_factor) );
-}
-
 bool register_keccakc_algo( algo_gate_t* gate )
 {
   gate->optimizations = AVX2_OPT;
   gate->gen_merkle_root = (void*)&sha256d_gen_merkle_root;
-  gate->set_target      = (void*)&keccakc_set_target;
-  gate->get_max64       = (void*)&keccak_get_max64;
+  opt_target_factor = 256.0;
 #if defined (KECCAK_4WAY)
   gate->scanhash  = (void*)&scanhash_keccak_4way;
   gate->hash      = (void*)&keccakhash_4way;

algo/keccak/keccak-hash-4way.c

@@ -370,18 +370,23 @@ static const sph_u64 RC[] = {
 
 static void keccak64_init( keccak64_ctx_m256i *kc, unsigned out_size )
 {
-   int i;
-   for (i = 0; i < 25; i ++)
-          kc->w[i] = _mm256_setzero_si256();
+   __m256i zero = m256_zero;
+   __m256i neg1 = m256_neg1;
 
    // Initialization for the "lane complement".
-   kc->w[ 1] = m256_neg1;
-   kc->w[ 2] = m256_neg1;
-   kc->w[ 8] = m256_neg1;
-   kc->w[12] = m256_neg1;
-   kc->w[17] = m256_neg1;
-   kc->w[20] = m256_neg1;
-   kc->ptr = 0;
+   kc->w[ 0] = zero;   kc->w[ 1] = neg1;
+   kc->w[ 2] = neg1;   kc->w[ 3] = zero;
+   kc->w[ 4] = zero;   kc->w[ 5] = zero;
+   kc->w[ 6] = zero;   kc->w[ 7] = zero;
+   kc->w[ 8] = neg1;   kc->w[ 9] = zero;
+   kc->w[10] = zero;   kc->w[11] = zero;
+   kc->w[12] = neg1;   kc->w[13] = zero;
+   kc->w[14] = zero;   kc->w[15] = zero;
+   kc->w[16] = zero;   kc->w[17] = neg1;
+   kc->w[18] = zero;   kc->w[19] = zero;
+   kc->w[20] = neg1;   kc->w[21] = zero;
+   kc->w[22] = zero;   kc->w[23] = zero;
+   kc->w[24] = zero;   kc->ptr = 0;
    kc->lim = 200 - (out_size >> 2);
 }
 
@@ -441,8 +446,8 @@ static void keccak64_close( keccak64_ctx_m256i *kc, void *dst, size_t byte_len,
     eb = 0x100  >> 8;
     if ( kc->ptr == (lim - 8) )
     {
-        uint64_t t = eb | 0x8000000000000000;
-        u.tmp[0] = _mm256_set_epi64x( t, t, t, t );
+        const uint64_t t = eb | 0x8000000000000000;
+        u.tmp[0] = m256_const1_64( t );
         j = 8;
     }
     else
@@ -450,8 +455,7 @@ static void keccak64_close( keccak64_ctx_m256i *kc, void *dst, size_t byte_len,
         j = lim - kc->ptr;
         u.tmp[0] = _mm256_set_epi64x( eb, eb, eb, eb );
         memset_zero_256( u.tmp + 1, (j>>3) - 2 );
-        u.tmp[ (j>>3) - 1] = _mm256_set_epi64x( 0x8000000000000000,
-                0x8000000000000000, 0x8000000000000000, 0x8000000000000000);
+        u.tmp[ (j>>3) - 1] = m256_const1_64( 0x8000000000000000 );
     }
     keccak64_core( kc, u.tmp, j, lim );
     /* Finalize the "lane complement" */
@@ -461,9 +465,7 @@ static void keccak64_close( keccak64_ctx_m256i *kc, void *dst, size_t byte_len,
     NOT64( kc->w[12], kc->w[12] );
     NOT64( kc->w[17], kc->w[17] );
     NOT64( kc->w[20], kc->w[20] );
-    for ( j = 0; j < m256_len; j++ )
-         u.tmp[j] =  kc->w[j]; 
-    memcpy_256( dst, u.tmp, m256_len );
+    memcpy_256( dst, kc->w, m256_len );
 }
 
 void keccak256_4way_init( void *kc )

algo/lanehash/lane.h

@@ -0,0 +1,50 @@
+/*
+ * Copyright (c) 2008 Sebastiaan Indesteege
+ *                              <sebastiaan.indesteege@esat.kuleuven.be>
+ *
+ * Permission to use, copy, modify, and distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ */
+
+/*
+ * Optimised ANSI-C implementation of LANE
+ */
+
+#ifndef LANE_H
+#define LANE_H
+
+#include <string.h>
+//#include "algo/sha/sha3-defs.h"
+#include <stdint.h>
+
+typedef unsigned char BitSequence;
+typedef unsigned long long DataLength;
+
+//typedef enum { SUCCESS = 0, FAIL = 1, BAD_HASHBITLEN = 2, BAD_DATABITLEN = 3 } HashReturn;
+
+//typedef unsigned char u8;
+//typedef unsigned int u32;
+//typedef unsigned long long u64;
+
+typedef struct {
+  int hashbitlen;
+  uint64_t ctr;
+  uint32_t h[16];
+  uint8_t buffer[128];
+} hashState;
+
+void laneInit (hashState *state, int hashbitlen);
+void laneUpdate (hashState *state, const BitSequence *data, DataLength databitlen);
+void laneFinal (hashState *state, BitSequence *hashval);
+void laneHash (int hashbitlen, const BitSequence *data, DataLength databitlen, BitSequence *hashval);
+
+#endif /* LANE_H */

algo/luffa/luffa-hash-2way.c

@@ -1,23 +1,3 @@
-/*
- * luffa_for_sse2.c
- * Version 2.0 (Sep 15th 2009)
- *
- * Copyright (C) 2008-2009 Hitachi, Ltd. All rights reserved.
- *
- * Hitachi, Ltd. is the owner of this software and hereby grant
- * the U.S. Government and any interested party the right to use
- * this software for the purposes of the SHA-3 evaluation process,
- * notwithstanding that this software is copyrighted.
- *
- * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
- * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
- * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
- * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
- * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
- * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
- * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
- */
-
 #include <string.h>
 #include <immintrin.h>
 #include "luffa-hash-2way.h"
@@ -26,31 +6,30 @@
 
 #include "simd-utils.h"
 
-#define MASK _mm256_set_epi32( 0UL, 0UL, 0UL, 0xffffffffUL, \
-                               0UL, 0UL, 0UL, 0xffffffffUL )
+#define cns(i)  m256_const1_128( ( (__m128i*)CNS_INIT)[i] )
 
 #define ADD_CONSTANT(a,b,c0,c1)\
     a = _mm256_xor_si256(a,c0);\
     b = _mm256_xor_si256(b,c1);\
 
-#define MULT2(a0,a1) \
+#define MULT2( a0, a1, mask ) \
 do { \
-  register __m256i b = _mm256_xor_si256( a0, \
-                   _mm256_shuffle_epi32( _mm256_and_si256(a1,MASK), 16 ) ); \
+  __m256i b = _mm256_xor_si256( a0, \
+                   _mm256_shuffle_epi32( _mm256_and_si256(a1,mask), 16 ) ); \
   a0 = _mm256_or_si256( _mm256_srli_si256(b,4), _mm256_slli_si256(a1,12) ); \
   a1 = _mm256_or_si256( _mm256_srli_si256(a1,4), _mm256_slli_si256(b,12) );  \
 } while(0)
 
 // confirm pointer arithmetic
 // ok but use array indexes
-#define STEP_PART(x,c,t)\
+#define STEP_PART(x,c0,c1,t)\
     SUBCRUMB(*x,*(x+1),*(x+2),*(x+3),*t);\
     SUBCRUMB(*(x+5),*(x+6),*(x+7),*(x+4),*t);\
     MIXWORD(*x,*(x+4),*t,*(t+1));\
     MIXWORD(*(x+1),*(x+5),*t,*(t+1));\
     MIXWORD(*(x+2),*(x+6),*t,*(t+1));\
     MIXWORD(*(x+3),*(x+7),*t,*(t+1));\
-    ADD_CONSTANT(*x, *(x+4), *c, *(c+1));
+    ADD_CONSTANT(*x, *(x+4), c0, c1);
 
 #define SUBCRUMB(a0,a1,a2,a3,t)\
     t  = _mm256_load_si256(&a0);\
@@ -245,7 +224,7 @@ static const uint32 CNS_INIT[128] __attribute((aligned(32))) = {
     0x00000000,0x00000000,0x00000000,0xfc053c31
 };
 
-__m256i CNS[32];
+
 
 /***************************************************/
 /* Round function         */
@@ -258,6 +237,7 @@ void rnd512_2way( luffa_2way_context *state, __m256i *msg )
     __m256i msg0, msg1;
     __m256i tmp[2];
     __m256i x[8];
+    const __m256i MASK = m256_const2_64( 0, 0x00000000ffffffff );
 
     t0 = chainv[0];
     t1 = chainv[1];
@@ -271,7 +251,7 @@ void rnd512_2way( luffa_2way_context *state, __m256i *msg )
     t0 = _mm256_xor_si256( t0, chainv[8] );
     t1 = _mm256_xor_si256( t1, chainv[9] );
 
-    MULT2( t0, t1 );
+    MULT2( t0, t1, MASK );
 
     msg0 = _mm256_shuffle_epi32( msg[0], 27 );
     msg1 = _mm256_shuffle_epi32( msg[1], 27 );
@@ -290,66 +270,66 @@ void rnd512_2way( luffa_2way_context *state, __m256i *msg )
     t0 = chainv[0];
     t1 = chainv[1];
 
-    MULT2( chainv[0], chainv[1]);
+    MULT2( chainv[0], chainv[1], MASK );
     chainv[0] = _mm256_xor_si256( chainv[0], chainv[2] );
     chainv[1] = _mm256_xor_si256( chainv[1], chainv[3] );
 
-    MULT2( chainv[2], chainv[3]);
+    MULT2( chainv[2], chainv[3], MASK );
     chainv[2] = _mm256_xor_si256(chainv[2], chainv[4]);
     chainv[3] = _mm256_xor_si256(chainv[3], chainv[5]);
 
-    MULT2( chainv[4], chainv[5]);
+    MULT2( chainv[4], chainv[5], MASK );
     chainv[4] = _mm256_xor_si256(chainv[4], chainv[6]);
     chainv[5] = _mm256_xor_si256(chainv[5], chainv[7]);
 
-    MULT2( chainv[6], chainv[7]);
+    MULT2( chainv[6], chainv[7], MASK );
     chainv[6] = _mm256_xor_si256(chainv[6], chainv[8]);
     chainv[7] = _mm256_xor_si256(chainv[7], chainv[9]);
 
-    MULT2( chainv[8], chainv[9]);
+    MULT2( chainv[8], chainv[9], MASK );
     chainv[8] = _mm256_xor_si256( chainv[8], t0 );
     chainv[9] = _mm256_xor_si256( chainv[9], t1 );
 
     t0 = chainv[8];
     t1 = chainv[9];
 
-    MULT2( chainv[8], chainv[9]);
+    MULT2( chainv[8], chainv[9], MASK );
     chainv[8] = _mm256_xor_si256( chainv[8], chainv[6] );
     chainv[9] = _mm256_xor_si256( chainv[9], chainv[7] );
 
-    MULT2( chainv[6], chainv[7]);
+    MULT2( chainv[6], chainv[7], MASK );
     chainv[6] = _mm256_xor_si256( chainv[6], chainv[4] );
     chainv[7] = _mm256_xor_si256( chainv[7], chainv[5] );
 
-    MULT2( chainv[4], chainv[5]);
+    MULT2( chainv[4], chainv[5], MASK );
     chainv[4] = _mm256_xor_si256( chainv[4], chainv[2] );
     chainv[5] = _mm256_xor_si256( chainv[5], chainv[3] );
 
-    MULT2( chainv[2], chainv[3] );
+    MULT2( chainv[2], chainv[3], MASK );
     chainv[2] = _mm256_xor_si256( chainv[2], chainv[0] );
     chainv[3] = _mm256_xor_si256( chainv[3], chainv[1] );
 
-    MULT2( chainv[0], chainv[1] );
+    MULT2( chainv[0], chainv[1], MASK );
     chainv[0] = _mm256_xor_si256( _mm256_xor_si256( chainv[0], t0 ), msg0 );
     chainv[1] = _mm256_xor_si256( _mm256_xor_si256( chainv[1], t1 ), msg1 );
 
-    MULT2( msg0, msg1);
+    MULT2( msg0, msg1, MASK );
     chainv[2] = _mm256_xor_si256( chainv[2], msg0 );
     chainv[3] = _mm256_xor_si256( chainv[3], msg1 );
 
-    MULT2( msg0, msg1);
+    MULT2( msg0, msg1, MASK );
     chainv[4] = _mm256_xor_si256( chainv[4], msg0 );
     chainv[5] = _mm256_xor_si256( chainv[5], msg1 );
 
-    MULT2( msg0, msg1);
+    MULT2( msg0, msg1, MASK );
     chainv[6] = _mm256_xor_si256( chainv[6], msg0 );
     chainv[7] = _mm256_xor_si256( chainv[7], msg1 );
 
-    MULT2( msg0, msg1);
+    MULT2( msg0, msg1, MASK );
     chainv[8] = _mm256_xor_si256( chainv[8], msg0 );
     chainv[9] = _mm256_xor_si256( chainv[9], msg1 );
 
-    MULT2( msg0, msg1);
+    MULT2( msg0, msg1, MASK );
 
     chainv[3] = _mm256_or_si256( _mm256_slli_epi32( chainv[3],  1 ),
                                  _mm256_srli_epi32( chainv[3], 31 ) );
@@ -365,14 +345,14 @@ void rnd512_2way( luffa_2way_context *state, __m256i *msg )
                 chainv[1],chainv[3],chainv[5],chainv[7],
                 x[4], x[5], x[6], x[7] );
 
-    STEP_PART( &x[0], &CNS[ 0], &tmp[0] );
-    STEP_PART( &x[0], &CNS[ 2], &tmp[0] );
-    STEP_PART( &x[0], &CNS[ 4], &tmp[0] );
-    STEP_PART( &x[0], &CNS[ 6], &tmp[0] );
-    STEP_PART( &x[0], &CNS[ 8], &tmp[0] );
-    STEP_PART( &x[0], &CNS[10], &tmp[0] );
-    STEP_PART( &x[0], &CNS[12], &tmp[0] );
-    STEP_PART( &x[0], &CNS[14], &tmp[0] );
+    STEP_PART( &x[0], cns( 0), cns( 1), &tmp[0] );
+    STEP_PART( &x[0], cns( 2), cns( 3), &tmp[0] );
+    STEP_PART( &x[0], cns( 4), cns( 5), &tmp[0] );
+    STEP_PART( &x[0], cns( 6), cns( 7), &tmp[0] );
+    STEP_PART( &x[0], cns( 8), cns( 9), &tmp[0] );
+    STEP_PART( &x[0], cns(10), cns(11), &tmp[0] );
+    STEP_PART( &x[0], cns(12), cns(13), &tmp[0] );
+    STEP_PART( &x[0], cns(14), cns(15), &tmp[0] );
 
     MIXTON1024( x[0], x[1], x[2], x[3],
                 chainv[0], chainv[2], chainv[4],chainv[6],
@@ -380,25 +360,24 @@ void rnd512_2way( luffa_2way_context *state, __m256i *msg )
                 chainv[1],chainv[3],chainv[5],chainv[7]);
 
     /* Process last 256-bit block */
-    STEP_PART2( chainv[8], chainv[9], t0, t1, CNS[16], CNS[17],
+    STEP_PART2( chainv[8], chainv[9], t0, t1, cns(16), cns(17),
                 tmp[0], tmp[1] );
-    STEP_PART2( chainv[8], chainv[9], t0, t1, CNS[18], CNS[19],
+    STEP_PART2( chainv[8], chainv[9], t0, t1, cns(18), cns(19),
                 tmp[0], tmp[1] );
-    STEP_PART2( chainv[8], chainv[9], t0, t1, CNS[20], CNS[21],
+    STEP_PART2( chainv[8], chainv[9], t0, t1, cns(20), cns(21),
                 tmp[0], tmp[1] );
-    STEP_PART2( chainv[8], chainv[9], t0, t1, CNS[22], CNS[23],
+    STEP_PART2( chainv[8], chainv[9], t0, t1, cns(22), cns(23),
                 tmp[0], tmp[1] );
-    STEP_PART2( chainv[8], chainv[9], t0, t1, CNS[24], CNS[25],
+    STEP_PART2( chainv[8], chainv[9], t0, t1, cns(24), cns(25),
                 tmp[0], tmp[1] );
-    STEP_PART2( chainv[8], chainv[9], t0, t1, CNS[26], CNS[27],
+    STEP_PART2( chainv[8], chainv[9], t0, t1, cns(26), cns(27),
                 tmp[0], tmp[1] );
-    STEP_PART2( chainv[8], chainv[9], t0, t1, CNS[28], CNS[29],
+    STEP_PART2( chainv[8], chainv[9], t0, t1, cns(28), cns(29),
                 tmp[0], tmp[1] );
-    STEP_PART2( chainv[8], chainv[9], t0, t1, CNS[30], CNS[31],
+    STEP_PART2( chainv[8], chainv[9], t0, t1, cns(30), cns(31),
                 tmp[0], tmp[1] );
 }
 
-
 /***************************************************/
 /* Finalization function  */
 /* state: hash context    */
@@ -410,8 +389,9 @@ void finalization512_2way( luffa_2way_context *state, uint32 *b )
     __m256i* chainv = state->chainv;
     __m256i t[2];
     __m256i zero[2];
-    zero[0] = zero[1] = _mm256_setzero_si256();
-
+    zero[0] = zero[1] = m256_zero;
+    const __m256i shuff_bswap32 = m256_const2_64( 0x0c0d0e0f08090a0b,
+                                                  0x0405060700010203 );
     /*---- blank round with m=0 ----*/
     rnd512_2way( state, zero );
 
@@ -433,8 +413,10 @@ void finalization512_2way( luffa_2way_context *state, uint32 *b )
     _mm256_store_si256( (__m256i*)&hash[0], t[0] );
     _mm256_store_si256( (__m256i*)&hash[8], t[1] );
 
-    casti_m256i( b, 0 ) = mm256_bswap_32( casti_m256i( hash, 0 ) );
-    casti_m256i( b, 1 ) = mm256_bswap_32( casti_m256i( hash, 1 ) );
+    casti_m256i( b, 0 ) = _mm256_shuffle_epi8(
+                                  casti_m256i( hash, 0 ), shuff_bswap32 );
+    casti_m256i( b, 1 ) = _mm256_shuffle_epi8( 
+                                  casti_m256i( hash, 1 ), shuff_bswap32 );
 
     rnd512_2way( state, zero );
 
@@ -455,26 +437,27 @@ void finalization512_2way( luffa_2way_context *state, uint32 *b )
     _mm256_store_si256( (__m256i*)&hash[0], t[0] );
     _mm256_store_si256( (__m256i*)&hash[8], t[1] );
 
-    casti_m256i( b, 2 ) = mm256_bswap_32( casti_m256i( hash, 0 ) );
-    casti_m256i( b, 3 ) = mm256_bswap_32( casti_m256i( hash, 1 ) );
+    casti_m256i( b, 2 ) = _mm256_shuffle_epi8( 
+                                  casti_m256i( hash, 0 ), shuff_bswap32 );
+    casti_m256i( b, 3 ) = _mm256_shuffle_epi8( 
+                                  casti_m256i( hash, 1 ), shuff_bswap32 );
 }
 
 int luffa_2way_init( luffa_2way_context *state, int hashbitlen )
 {
-    int i;
     state->hashbitlen = hashbitlen;
-
-    for ( i=0; i<32; i++ ) CNS[i] =
-          _mm256_set_epi32( CNS_INIT[ (i<<2) + 3 ], CNS_INIT[ (i<<2) +2 ],
-                            CNS_INIT[ (i<<2) + 1 ], CNS_INIT[ (i<<2)    ],
-                            CNS_INIT[ (i<<2) + 3 ], CNS_INIT[ (i<<2) +2 ],
-                            CNS_INIT[ (i<<2) + 1 ], CNS_INIT[ (i<<2)    ] );
-
-    for ( i=0; i<10; i++ ) state->chainv[i] =
-          _mm256_set_epi32( IV[ (i<<2) +3 ], IV[ (i<<2) +2 ],
-                            IV[ (i<<2) +1 ], IV[ (i<<2)    ],
-                            IV[ (i<<2) +3 ], IV[ (i<<2) +2 ],
-                            IV[ (i<<2) +1 ], IV[ (i<<2)    ] );
+    __m128i *iv = (__m128i*)IV;
+    
+    state->chainv[0] = m256_const1_128( iv[0] );
+    state->chainv[1] = m256_const1_128( iv[1] );
+    state->chainv[2] = m256_const1_128( iv[2] );
+    state->chainv[3] = m256_const1_128( iv[3] );
+    state->chainv[4] = m256_const1_128( iv[4] );
+    state->chainv[5] = m256_const1_128( iv[5] );
+    state->chainv[6] = m256_const1_128( iv[6] );
+    state->chainv[7] = m256_const1_128( iv[7] );
+    state->chainv[8] = m256_const1_128( iv[8] );
+    state->chainv[9] = m256_const1_128( iv[9] );
 
     ((__m256i*)state->buffer)[0] = m256_zero;
     ((__m256i*)state->buffer)[1] = m256_zero;
@@ -492,13 +475,15 @@ int luffa_2way_update( luffa_2way_context *state, const void *data,
     __m256i msg[2];
     int i;
     int blocks = (int)len >> 5;
+    const __m256i shuff_bswap32 = m256_const2_64( 0x0c0d0e0f08090a0b,
+                                                  0x0405060700010203 );
     state-> rembytes = (int)len & 0x1F;
 
     // full blocks
     for ( i = 0; i < blocks; i++, vdata+=2 )
     {
-       msg[0] = mm256_bswap_32( vdata[ 0] );
-       msg[1] = mm256_bswap_32( vdata[ 1 ] );
+       msg[0] = _mm256_shuffle_epi8( vdata[ 0 ], shuff_bswap32 );
+       msg[1] = _mm256_shuffle_epi8( vdata[ 1 ], shuff_bswap32 );
        rnd512_2way( state, msg );
     }
 
@@ -507,9 +492,8 @@ int luffa_2way_update( luffa_2way_context *state, const void *data,
     if ( state->rembytes  )
     {
       // remaining data bytes
-      buffer[0] = mm256_bswap_32( vdata[0] );
-      buffer[1] = _mm256_set_epi8( 0,0,0,0, 0,0,0,0, 0,0,0,0, 0x80,0,0,0,
-                                   0,0,0,0, 0,0,0,0, 0,0,0,0, 0x80,0,0,0 );
+      buffer[0] = _mm256_shuffle_epi8( vdata[0], shuff_bswap32 );
+      buffer[1] = m256_const2_64( 0, 0x0000000080000000 );
     }
     return 0;
 }
@@ -525,8 +509,7 @@ int luffa_2way_close( luffa_2way_context *state, void *hashval )
       rnd512_2way( state, buffer );
     else
     {     // empty pad block, constant data
-      msg[0] = _mm256_set_epi8( 0,0,0,0, 0,0,0,0, 0,0,0,0, 0x80,0,0,0,
-                                0,0,0,0, 0,0,0,0, 0,0,0,0, 0x80,0,0,0 );
+      msg[0] = m256_const2_64( 0, 0x0000000080000000 );
       msg[1] = m256_zero;
       rnd512_2way( state, msg );
     }
@@ -545,13 +528,16 @@ int luffa_2way_update_close( luffa_2way_context *state,
     __m256i msg[2];
     int i;
     const int blocks = (int)( inlen >> 5 );
+    const __m256i shuff_bswap32 = m256_const2_64( 0x0c0d0e0f08090a0b,
+                                                  0x0405060700010203 );
+
     state->rembytes = inlen & 0x1F;
 
     // full blocks
     for ( i = 0; i < blocks; i++, vdata+=2 )
     {
-       msg[0] = mm256_bswap_32( vdata[ 0 ] );
-       msg[1] = mm256_bswap_32( vdata[ 1 ] );
+       msg[0] = _mm256_shuffle_epi8( vdata[ 0 ], shuff_bswap32 );
+       msg[1] = _mm256_shuffle_epi8( vdata[ 1 ], shuff_bswap32 );
        rnd512_2way( state, msg );
     }
 
@@ -559,16 +545,14 @@ int luffa_2way_update_close( luffa_2way_context *state,
     if ( state->rembytes  )
     {
        // padding of partial block
-       msg[0] = mm256_bswap_32( vdata[0] );
-       msg[1] = _mm256_set_epi8( 0,0,0,0, 0,0,0,0, 0,0,0,0, 0x80,0,0,0,
-                                 0,0,0,0, 0,0,0,0, 0,0,0,0, 0x80,0,0,0 );
+       msg[0] = _mm256_shuffle_epi8( vdata[ 0 ], shuff_bswap32 );
+       msg[1] = m256_const2_64( 0, 0x0000000080000000 );
        rnd512_2way( state, msg );
     }
     else
     {
        // empty pad block
-       msg[0] = _mm256_set_epi8( 0,0,0,0, 0,0,0,0, 0,0,0,0, 0x80,0,0,0,
-                                 0,0,0,0, 0,0,0,0, 0,0,0,0, 0x80,0,0,0 );
+       msg[0] = m256_const2_64( 0, 0x0000000080000000 );
        msg[1] = m256_zero;
        rnd512_2way( state, msg );
     }

algo/luffa/luffa_for_sse2.c

@@ -541,7 +541,9 @@ static void finalization512( hashState_luffa *state, uint32 *b )
     uint32   hash[8] __attribute((aligned(64)));
     __m256i* chainv = (__m256i*)state->chainv;
     __m256i  t;
-    const __m128i zero = _mm_setzero_si128();
+    const __m128i zero = m128_zero;
+    const __m256i shuff_bswap32 = m256_const2_64( 0x0c0d0e0f08090a0b,
+                                                  0x0405060700010203 );
 
     rnd512( state, zero, zero );
 
@@ -555,7 +557,9 @@ static void finalization512( hashState_luffa *state, uint32 *b )
 
     _mm256_store_si256( (__m256i*)hash, t );
 
-    casti_m256i( b, 0 ) = mm256_bswap_32( casti_m256i( hash, 0 ) );
+    casti_m256i( b, 0 ) = _mm256_shuffle_epi8(
+                                 casti_m256i( hash, 0 ), shuff_bswap32 );
+//    casti_m256i( b, 0 ) = mm256_bswap_32( casti_m256i( hash, 0 ) );
 
     rnd512( state, zero, zero );
 
@@ -568,7 +572,9 @@ static void finalization512( hashState_luffa *state, uint32 *b )
 
     _mm256_store_si256( (__m256i*)hash, t );
 
-    casti_m256i( b, 1 ) = mm256_bswap_32( casti_m256i( hash, 0 ) );
+    casti_m256i( b, 1 ) = _mm256_shuffle_epi8( 
+                                 casti_m256i( hash, 0 ), shuff_bswap32 );
+//    casti_m256i( b, 1 ) = mm256_bswap_32( casti_m256i( hash, 0 ) );
 }
 
 #else

algo/lyra2/lyra2-gate.c

@@ -71,7 +71,7 @@ bool register_lyra2rev3_algo( algo_gate_t* gate )
 #endif
   gate->optimizations = SSE2_OPT | SSE42_OPT | AVX2_OPT;
   gate->miner_thread_init = (void*)&lyra2rev3_thread_init;
-  gate->set_target        = (void*)&alt_set_target;
+  opt_target_factor = 256.0;
   return true;
 };
 
@@ -105,7 +105,7 @@ bool register_lyra2rev2_algo( algo_gate_t* gate )
 #endif
   gate->optimizations = SSE2_OPT | AES_OPT | SSE42_OPT | AVX2_OPT;
   gate->miner_thread_init = (void*)&lyra2rev2_thread_init;
-  gate->set_target        = (void*)&alt_set_target;
+  opt_target_factor = 256.0;
   return true;
 };
 
@@ -127,8 +127,7 @@ bool register_lyra2z_algo( algo_gate_t* gate )
   gate->hash       = (void*)&lyra2z_hash;
 #endif
   gate->optimizations = SSE42_OPT | AVX2_OPT;
-  gate->get_max64  = (void*)&get_max64_0xffffLL;
-  gate->set_target = (void*)&alt_set_target;
+  opt_target_factor = 256.0;
   return true;
 };
 
@@ -147,15 +146,12 @@ bool register_lyra2h_algo( algo_gate_t* gate )
   gate->hash       = (void*)&lyra2h_hash;
 #endif
   gate->optimizations = SSE42_OPT | AVX2_OPT;
-  gate->get_max64  = (void*)&get_max64_0xffffLL;
-  gate->set_target = (void*)&alt_set_target;
+  opt_target_factor = 256.0;
   return true;
 };
 
 /////////////////////////////////
 
-int64_t allium_get_max64_0xFFFFLL() { return 0xFFFFLL; }
-
 bool register_allium_algo( algo_gate_t* gate )
 {
 #if defined (ALLIUM_4WAY)
@@ -168,8 +164,7 @@ bool register_allium_algo( algo_gate_t* gate )
   gate->hash      = (void*)&allium_hash;
 #endif
   gate->optimizations = SSE2_OPT | AES_OPT | SSE42_OPT | AVX2_OPT;
-  gate->set_target        = (void*)&alt_set_target;
-  gate->get_max64         = (void*)&allium_get_max64_0xFFFFLL;
+  opt_target_factor = 256.0;
   return true;
 };
 
@@ -182,6 +177,7 @@ int phi2_get_work_data_size() { return phi2_use_roots ? 144 : 128; }
 
 void phi2_decode_extra_data( struct work *work )
 {
+   phi2_use_roots = false;
    if ( work->data[0] & ( 1<<30 ) ) phi2_use_roots = true;
    else for ( int i = 20; i < 36; i++ )
    {
@@ -213,8 +209,7 @@ bool register_phi2_algo( algo_gate_t* gate )
    gate->get_work_data_size = (void*)&phi2_get_work_data_size;
    gate->decode_extra_data  = (void*)&phi2_decode_extra_data;
    gate->build_extraheader  = (void*)&phi2_build_extraheader;
-   gate->set_target         = (void*)&alt_set_target; 
-   gate->get_max64          = (void*)&get_max64_0xffffLL;
+   opt_target_factor = 256.0;
 #if defined(PHI2_4WAY)
    gate->scanhash           = (void*)&scanhash_phi2_4way;
 #else

algo/lyra2/lyra2.c

@@ -60,7 +60,7 @@ int LYRA2REV2( uint64_t* wholeMatrix, void *K, uint64_t kLen, const void *pwd,
    int64_t step = 1; //Visitation step (used during Setup and Wandering phases)
    int64_t window = 2; //Visitation window (used to define which rows can be revisited during Setup)
    int64_t gap = 1; //Modifier to the step, assuming the values 1 or -1
-   int64_t i; //auxiliary iteration counter
+//   int64_t i; //auxiliary iteration counter
    int64_t v64; // 64bit var for memcpy
    //====================================================================/
 
@@ -128,17 +128,22 @@ int LYRA2REV2( uint64_t* wholeMatrix, void *K, uint64_t kLen, const void *pwd,
    //================= Initializing the Sponge State ====================//
    //Sponge state: 16 uint64_t, BLOCK_LEN_INT64 words of them for the bitrate (b) and the remainder for the capacity (c)
 
-   initState( state );
+//   initState( state );
 
    //========================= Setup Phase =============================//
    //Absorbing salt, password and basil: this is the only place in which the block length is hard-coded to 512 bits
    
    ptrWord = wholeMatrix;
+
+   absorbBlockBlake2Safe( state, ptrWord, nBlocksInput, BLOCK_LEN );
+/*
    for (i = 0; i < nBlocksInput; i++)
    {
        absorbBlockBlake2Safe( state, ptrWord ); //absorbs each block of pad(pwd || salt || basil)
        ptrWord += BLOCK_LEN; //goes to next block of pad(pwd || salt || basil)
    }
+*/
+
    //Initializes M[0] and M[1]
    reducedSqueezeRow0( state, &wholeMatrix[0], nCols ); //The locally copied password is most likely overwritten here
 
@@ -227,7 +232,7 @@ int LYRA2REV3( uint64_t* wholeMatrix, void *K, uint64_t kLen, const void *pwd,
    int64_t step = 1; //Visitation step (used during Setup and Wandering phases)
    int64_t window = 2; //Visitation window (used to define which rows can be revisited during Setup)
    int64_t gap = 1; //Modifier to the step, assuming the values 1 or -1
-   int64_t i; //auxiliary iteration counter
+//   int64_t i; //auxiliary iteration counter
    int64_t v64; // 64bit var for memcpy
    uint64_t instance = 0;
    //====================================================================/
@@ -302,17 +307,21 @@ int LYRA2REV3( uint64_t* wholeMatrix, void *K, uint64_t kLen, const void *pwd,
    //================= Initializing the Sponge State ====================//
    //Sponge state: 16 uint64_t, BLOCK_LEN_INT64 words of them for the bitrate (b) and the remainder for the capacity (c)
 
-   initState( state );
+//   initState( state );
 
    //========================= Setup Phase =============================//
    //Absorbing salt, password and basil: this is the only place in which the block length is hard-coded to 512 bits
 
    ptrWord = wholeMatrix;
+
+   absorbBlockBlake2Safe( state, ptrWord, nBlocksInput, BLOCK_LEN );
+/*
    for (i = 0; i < nBlocksInput; i++)
    {
        absorbBlockBlake2Safe( state, ptrWord ); //absorbs each block of pad(pwd || salt || basil)
        ptrWord += BLOCK_LEN; //goes to next block of pad(pwd || salt || basil)
    }
+*/
    //Initializes M[0] and M[1]
    reducedSqueezeRow0( state, &wholeMatrix[0], nCols ); //The locally copied password is most likely overwritten here
 
@@ -405,7 +414,7 @@ int LYRA2Z( uint64_t* wholeMatrix, void *K, uint64_t kLen, const void *pwd,
     int64_t step = 1; //Visitation step (used during Setup and Wandering phases)
     int64_t window = 2; //Visitation window (used to define which rows can be revisited during Setup)
     int64_t gap = 1; //Modifier to the step, assuming the values 1 or -1
-    int64_t i; //auxiliary iteration counter
+//    int64_t i; //auxiliary iteration counter
     //=======================================================================/
 
     //======= Initializing the Memory Matrix and pointers to it =============//
@@ -459,17 +468,21 @@ int LYRA2Z( uint64_t* wholeMatrix, void *K, uint64_t kLen, const void *pwd,
 //        if (state == NULL) {
 //                return -1;
 //        }
-    initState( state );
+//    initState( state );
 
     //============================== Setup Phase =============================//
     //Absorbing salt, password and basil: this is the only place in which the block length is hard-coded to 512 bits
-        uint64_t *ptrWord = wholeMatrix;
+    uint64_t *ptrWord = wholeMatrix;
+
+    absorbBlockBlake2Safe( state, ptrWord, nBlocksInput,
+                           BLOCK_LEN_BLAKE2_SAFE_INT64 );
+/*
     for ( i = 0; i < nBlocksInput; i++ )
     {
       absorbBlockBlake2Safe( state, ptrWord ); //absorbs each block of pad(pwd || salt || basil)
       ptrWord += BLOCK_LEN_BLAKE2_SAFE_INT64; //goes to next block of pad(pwd || salt || basil)
     }
-
+*/
     //Initializes M[0] and M[1]
         reducedSqueezeRow0(state, &wholeMatrix[0], nCols); //The locally copied password is most likely overwritten here
         reducedDuplexRow1(state, &wholeMatrix[0], &wholeMatrix[ROW_LEN_INT64], nCols);
@@ -623,17 +636,21 @@ int LYRA2RE( void *K, uint64_t kLen, const void *pwd, const uint64_t pwdlen,
    //================= Initializing the Sponge State ====================//
    //Sponge state: 16 uint64_t, BLOCK_LEN_INT64 words of them for the bitrate (b) and the remainder for the capacity (c)
 
-   initState( state );
+//   initState( state );
 
    //========================= Setup Phase =============================//
    //Absorbing salt, password and basil: this is the only place in which the block length is hard-coded to 512 bits
 
    ptrWord = wholeMatrix;
+
+   absorbBlockBlake2Safe( state, ptrWord, nBlocksInput, BLOCK_LEN );
+/*
    for (i = 0; i < nBlocksInput; i++)
    {
        absorbBlockBlake2Safe( state, ptrWord ); //absorbs each block of pad(pwd || salt || basil)
        ptrWord += BLOCK_LEN; //goes to next block of pad(pwd || salt || basil)
    }
+*/
    //Initializes M[0] and M[1]
    reducedSqueezeRow0( state, &wholeMatrix[0], nCols ); //The locally copied password is most likely overwritten here
 

algo/lyra2/lyra2re.c

@@ -113,24 +113,13 @@ int scanhash_lyra2re( struct work *work, uint32_t max_nonce,
 	return 0;
 }
 
-int64_t lyra2re_get_max64 ()
-{
-  return 0xffffLL;
-}
-
-void lyra2re_set_target ( struct work* work, double job_diff )
-{
-   work_set_target(work, job_diff / (128.0 * opt_diff_factor) );
-}
-
 bool register_lyra2re_algo( algo_gate_t* gate )
 {
   init_lyra2re_ctx();
   gate->optimizations = SSE2_OPT | AES_OPT | SSE42_OPT | AVX2_OPT;
   gate->scanhash   = (void*)&scanhash_lyra2re;
   gate->hash       = (void*)&lyra2re_hash;
-  gate->get_max64  = (void*)&lyra2re_get_max64;
-  gate->set_target = (void*)&lyra2re_set_target;
+  opt_target_factor = 128.0;
   return true;
 };
 

algo/lyra2/lyra2rev3-4way.c

@@ -86,7 +86,7 @@ void lyra2rev3_8way_hash( void *state, const void *input )
 
    }
 
-int scanhash_lyra2rev3_8way( struct work *work, uint32_t max_nonce,
+int scanhash_lyra2rev3_8way( struct work *work, const uint32_t max_nonce,
                              uint64_t *hashes_done, struct thr_info *mythr )
 {
    uint32_t hash[8*8] __attribute__ ((aligned (64)));
@@ -94,12 +94,12 @@ int scanhash_lyra2rev3_8way( struct work *work, uint32_t max_nonce,
    uint32_t *hash7 = &(hash[7<<3]);
    uint32_t lane_hash[8] __attribute__ ((aligned (32)));
    uint32_t *pdata = work->data;
-   uint32_t *ptarget = work->target;
+   const uint32_t *ptarget = work->target;
    const uint32_t first_nonce = pdata[19];
    uint32_t n = first_nonce;
    const uint32_t Htarg = ptarget[7];
    __m256i  *noncev = (__m256i*)vdata + 19;   // aligned
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id;  // thr_id arg is deprecated
 
    if ( opt_benchmark )
       ( (uint32_t*)ptarget )[7] = 0x0000ff;
@@ -113,17 +113,18 @@ int scanhash_lyra2rev3_8way( struct work *work, uint32_t max_nonce,
       lyra2rev3_8way_hash( hash, vdata );
       pdata[19] = n;
 
-      for ( int lane = 0; lane < 8; lane++ ) if ( hash7[lane] <= Htarg )
+      for ( int lane = 0; lane < 8; lane++ )
+      if ( unlikely( hash7[lane] <= Htarg ) )
       {
          extr_lane_8x32( lane_hash, hash, lane, 256 );
-         if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
+         if ( likely( 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);
+   } while ( likely( (n < max_nonce-8) && !work_restart[thr_id].restart ) );
    *hashes_done = n - first_nonce + 1;
    return 0;
 }
@@ -186,7 +187,7 @@ void lyra2rev3_4way_hash( void *state, const void *input )
    bmw256_4way_close( &ctx.bmw, state );
 }
 
-int scanhash_lyra2rev3_4way( struct work *work, uint32_t max_nonce,
+int scanhash_lyra2rev3_4way( struct work *work, const uint32_t max_nonce,
                              uint64_t *hashes_done, struct thr_info *mythr ) 
 {
    uint32_t hash[8*4] __attribute__ ((aligned (64)));
@@ -194,12 +195,12 @@ int scanhash_lyra2rev3_4way( struct work *work, uint32_t max_nonce,
    uint32_t *hash7 = &(hash[7<<2]);
    uint32_t lane_hash[8] __attribute__ ((aligned (32)));
    uint32_t *pdata = work->data;
-   uint32_t *ptarget = work->target;
+   const uint32_t *ptarget = work->target;
    const uint32_t first_nonce = pdata[19];
    uint32_t n = first_nonce;
    const uint32_t Htarg = ptarget[7];
    __m128i  *noncev = (__m128i*)vdata + 19;   // aligned
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id;  // thr_id arg is deprecated
    
    if ( opt_benchmark )
       ( (uint32_t*)ptarget )[7] = 0x0000ff;

algo/lyra2/lyra2rev3.c

@@ -32,27 +32,27 @@ void l2v3_blake256_midstate( const void* input )
 
 void lyra2rev3_hash( void *state, const void *input )
 {
-        lyra2v3_ctx_holder ctx __attribute__ ((aligned (64))); 
-        memcpy( &ctx, &lyra2v3_ctx, sizeof(lyra2v3_ctx) );
-        uint8_t hash[128] __attribute__ ((aligned (64)));
-        #define hashA hash
-        #define hashB hash+64
-        const int midlen = 64;            // bytes
-        const int tail   = 80 - midlen;   // 16
+   lyra2v3_ctx_holder ctx __attribute__ ((aligned (64))); 
+   memcpy( &ctx, &lyra2v3_ctx, sizeof(lyra2v3_ctx) );
+   uint8_t hash[128] __attribute__ ((aligned (64)));
+   #define hashA hash
+   #define hashB hash+64
+   const int midlen = 64;            // bytes
+   const int tail   = 80 - midlen;   // 16
 
-        memcpy( &ctx.blake, &l2v3_blake_mid, sizeof l2v3_blake_mid );
-	sph_blake256( &ctx.blake, (uint8_t*)input + midlen, tail );
-	sph_blake256_close( &ctx.blake, hash );
+   memcpy( &ctx.blake, &l2v3_blake_mid, sizeof l2v3_blake_mid );
+   sph_blake256( &ctx.blake, (uint8_t*)input + midlen, tail );
+   sph_blake256_close( &ctx.blake, hash );
 
-        LYRA2REV3( l2v3_wholeMatrix, hash, 32, hash, 32, hash, 32, 1, 4, 4 );
+   LYRA2REV3( l2v3_wholeMatrix, hash, 32, hash, 32, hash, 32, 1, 4, 4 );
 
-        cubehashUpdateDigest( &ctx.cube, (byte*) hashA,
-                              (const byte*) hash, 32 );
+   cubehashUpdateDigest( &ctx.cube, (byte*) hashA,
+                         (const byte*) hash, 32 );
 
-	LYRA2REV3( l2v3_wholeMatrix, hash, 32, hash, 32, hash, 32, 1, 4, 4 );
+   LYRA2REV3( l2v3_wholeMatrix, hash, 32, hash, 32, hash, 32, 1, 4, 4 );
 
-	sph_bmw256( &ctx.bmw, hash, 32 );
-	sph_bmw256_close( &ctx.bmw, hash );
+   sph_bmw256( &ctx.bmw, hash, 32 );
+   sph_bmw256_close( &ctx.bmw, hash );
 
 	memcpy( state, hash, 32 );
 }

algo/lyra2/lyra2z330.c

@@ -53,11 +53,6 @@ int scanhash_lyra2z330( struct work *work, uint32_t max_nonce,
    return 0;
 }
 
-void lyra2z330_set_target( struct work* work, double job_diff )
-{
- work_set_target( work, job_diff / (256.0 * opt_diff_factor) );
-}
-
 bool lyra2z330_thread_init()
 {
    const int64_t ROW_LEN_INT64 = BLOCK_LEN_INT64 * 256; // nCols
@@ -75,8 +70,7 @@ bool register_lyra2z330_algo( algo_gate_t* gate )
   gate->miner_thread_init = (void*)&lyra2z330_thread_init;
   gate->scanhash   = (void*)&scanhash_lyra2z330;
   gate->hash       = (void*)&lyra2z330_hash;
-  gate->get_max64  = (void*)&get_max64_0xffffLL;
-  gate->set_target = (void*)&lyra2z330_set_target;
+  opt_target_factor = 256.0;
   return true;
 };
 

algo/lyra2/sponge.c

@@ -40,29 +40,32 @@
  */
 inline void initState( uint64_t State[/*16*/] )
 {
+
+   /*
 #if defined (__AVX2__)
 
   __m256i* state = (__m256i*)State;
- 
-  state[0] = _mm256_setzero_si256();
-  state[1] = _mm256_setzero_si256();
-  state[2] = _mm256_set_epi64x( blake2b_IV[3], blake2b_IV[2],
-                                blake2b_IV[1], blake2b_IV[0] );
-  state[3] = _mm256_set_epi64x( blake2b_IV[7], blake2b_IV[6],
-                                blake2b_IV[5], blake2b_IV[4] );
+  const __m256i zero = m256_zero; 
+  state[0] = zero;
+  state[1] = zero;
+  state[2] = m256_const_64( 0xa54ff53a5f1d36f1ULL, 0x3c6ef372fe94f82bULL,
+                            0xbb67ae8584caa73bULL, 0x6a09e667f3bcc908ULL );
+  state[3] = m256_const_64( 0x5be0cd19137e2179ULL, 0x1f83d9abfb41bd6bULL,
+                            0x9b05688c2b3e6c1fULL, 0x510e527fade682d1ULL );
 
 #elif defined (__SSE2__)
 
   __m128i* state = (__m128i*)State;
+  const __m128i zero = m128_zero;   
 
-  state[0] = _mm_setzero_si128();
-  state[1] = _mm_setzero_si128();
-  state[2] = _mm_setzero_si128();
-  state[3] = _mm_setzero_si128();
-  state[4] = _mm_set_epi64x( blake2b_IV[1], blake2b_IV[0] );
-  state[5] = _mm_set_epi64x( blake2b_IV[3], blake2b_IV[2] );
-  state[6] = _mm_set_epi64x( blake2b_IV[5], blake2b_IV[4] );
-  state[7] = _mm_set_epi64x( blake2b_IV[7], blake2b_IV[6] );
+  state[0] = zero;
+  state[1] = zero;
+  state[2] = zero;
+  state[3] = zero;
+  state[4] = m128_const_64( 0xbb67ae8584caa73bULL, 0x6a09e667f3bcc908ULL );
+  state[5] = m128_const_64( 0xa54ff53a5f1d36f1ULL, 0x3c6ef372fe94f82bULL );
+  state[6] = m128_const_64( 0x9b05688c2b3e6c1fULL, 0x510e527fade682d1ULL );
+  state[7] = m128_const_64( 0x5be0cd19137e2179ULL, 0x1f83d9abfb41bd6bULL );
 
 #else
     //First 512 bis are zeros
@@ -77,6 +80,8 @@ inline void initState( uint64_t State[/*16*/] )
     State[14] = blake2b_IV[6];
     State[15] = blake2b_IV[7];
 #endif
+*/
+
 }
 
 /**
@@ -250,43 +255,74 @@ inline void absorbBlock( uint64_t *State, const uint64_t *In )
  * @param state The current state of the sponge
  * @param in    The block to be absorbed (BLOCK_LEN_BLAKE2_SAFE_INT64 words)
  */
-inline void absorbBlockBlake2Safe( uint64_t *State, const uint64_t *In )
+inline void absorbBlockBlake2Safe( uint64_t *State, const uint64_t *In,
+                      const uint64_t nBlocks, const uint64_t block_len )
 {
-    //XORs the first BLOCK_LEN_BLAKE2_SAFE_INT64 words of "in" with the current state
+// XORs the first BLOCK_LEN_BLAKE2_SAFE_INT64 words of "in" with
+// the IV.
 #if defined (__AVX2__)
 
-    register __m256i state0, state1, state2, state3;
+  register __m256i state0, state1, state2, state3;
+
+  state0 = 
+  state1 = m256_zero;
+  state2 = m256_const_64( 0xa54ff53a5f1d36f1ULL, 0x3c6ef372fe94f82bULL,
+                          0xbb67ae8584caa73bULL, 0x6a09e667f3bcc908ULL );
+  state3 = m256_const_64( 0x5be0cd19137e2179ULL, 0x1f83d9abfb41bd6bULL,
+                          0x9b05688c2b3e6c1fULL, 0x510e527fade682d1ULL );
+
+  for ( int i = 0; i < nBlocks; i++ )
+  { 
     __m256i *in = (__m256i*)In;
-
-    state0 = _mm256_load_si256( (__m256i*)State     );
-    state1 = _mm256_load_si256( (__m256i*)State + 1 );
-    state2 = _mm256_load_si256( (__m256i*)State + 2 );
-    state3 = _mm256_load_si256( (__m256i*)State + 3 );
-
     state0 = _mm256_xor_si256( state0, in[0] );
     state1 = _mm256_xor_si256( state1, in[1] );
 
     LYRA_12_ROUNDS_AVX2( state0, state1, state2, state3 );
+    In += block_len;
+  }
 
-    _mm256_store_si256( (__m256i*)State,     state0 );
-    _mm256_store_si256( (__m256i*)State + 1, state1 );
-    _mm256_store_si256( (__m256i*)State + 2, state2 );
-    _mm256_store_si256( (__m256i*)State + 3, state3 );
+  _mm256_store_si256( (__m256i*)State,     state0 );
+  _mm256_store_si256( (__m256i*)State + 1, state1 );
+  _mm256_store_si256( (__m256i*)State + 2, state2 );
+  _mm256_store_si256( (__m256i*)State + 3, state3 );
 
 #elif defined (__SSE2__)
 
-    __m128i* state = (__m128i*)State;
+  __m128i state0, state1, state2, state3, state4, state5, state6, state7;
+
+  state0 = 
+  state1 =
+  state2 =
+  state3 = m128_zero;
+  state4 = m128_const_64( 0xbb67ae8584caa73bULL, 0x6a09e667f3bcc908ULL );
+  state5 = m128_const_64( 0xa54ff53a5f1d36f1ULL, 0x3c6ef372fe94f82bULL );
+  state6 = m128_const_64( 0x9b05688c2b3e6c1fULL, 0x510e527fade682d1ULL );
+  state7 = m128_const_64( 0x5be0cd19137e2179ULL, 0x1f83d9abfb41bd6bULL );
+
+  for ( int i = 0; i < nBlocks; i++ )
+  { 
     __m128i* in    = (__m128i*)In;
 
-    state[0] = _mm_xor_si128( state[0], in[0] );
-    state[1] = _mm_xor_si128( state[1], in[1] );
-    state[2] = _mm_xor_si128( state[2], in[2] );
-    state[3] = _mm_xor_si128( state[3], in[3] );
+    state0 = _mm_xor_si128( state0, in[0] );
+    state1 = _mm_xor_si128( state1, in[1] );
+    state2 = _mm_xor_si128( state2, in[2] );
+    state3 = _mm_xor_si128( state3, in[3] );
 
     //Applies the transformation f to the sponge's state
-    LYRA_12_ROUNDS_AVX( state[0], state[1], state[2], state[3],
-                        state[4], state[5], state[6], state[7] );
+    LYRA_12_ROUNDS_AVX( state0, state1, state2, state3,
+                        state4, state5, state6, state7 );
+    In += block_len;
+  }
 
+  _mm_store_si128( (__m128i*)State,     state0 );
+  _mm_store_si128( (__m128i*)State + 1, state1 );
+  _mm_store_si128( (__m128i*)State + 2, state2 );
+  _mm_store_si128( (__m128i*)State + 3, state3 );
+  _mm_store_si128( (__m128i*)State + 4, state4 );
+  _mm_store_si128( (__m128i*)State + 5, state5 );
+  _mm_store_si128( (__m128i*)State + 6, state6 );
+  _mm_store_si128( (__m128i*)State + 7, state7 );
+  
 #else
 
     State[0] ^= In[0];

algo/lyra2/sponge.h

@@ -170,7 +170,8 @@ void reducedSqueezeRow0(uint64_t* state, uint64_t* row, uint64_t nCols);
 
 //---- Absorbs
 void absorbBlock(uint64_t *state, const uint64_t *in);
-void absorbBlockBlake2Safe(uint64_t *state, const uint64_t *in);
+void absorbBlockBlake2Safe( uint64_t *state, const uint64_t *in,
+                            const uint64_t nBlocks, const uint64_t block_len );
 
 //---- Duplexes
 void reducedDuplexRow1(uint64_t *state, uint64_t *rowIn, uint64_t *rowOut, uint64_t nCols);

algo/m7m.c

@@ -19,100 +19,89 @@
 #define EPS1 DBL_EPSILON
 #define EPS2 3.0e-11
 
-inline double exp_n(double xt)
+inline double exp_n( double xt )
 {
-    if(xt < -700.0)
+    if ( xt < -700.0 )
         return 0;
-    else if(xt > 700.0)
+    else if ( xt > 700.0 )
         return 1e200;
-    else if(xt > -0.8e-8 && xt < 0.8e-8)
-        return (1.0 + xt);
+    else if ( xt > -0.8e-8 && xt < 0.8e-8 )
+        return ( 1.0 + xt );
     else
-        return exp(xt);
+        return exp( xt );
 }
 
-inline double exp_n2(double x1, double x2)
+inline double exp_n2( double x1, double x2 )
 {
-    double p1 = -700., p2 = -37., p3 = -0.8e-8, p4 = 0.8e-8, p5 = 37., p6 = 700.;
+    double p1 = -700., p2 = -37., p3 = -0.8e-8, p4 = 0.8e-8,
+           p5 = 37., p6 = 700.;
     double xt = x1 - x2;
-    if (xt < p1+1.e-200)
+    if ( xt < p1+1.e-200 )
         return 1.;
-    else if (xt > p1 && xt < p2 + 1.e-200)
+    else if ( xt > p1 && xt < p2 + 1.e-200 )
         return ( 1. - exp(xt) );
-    else if (xt > p2 && xt < p3 + 1.e-200)
-        return ( 1. / (1. + exp(xt)) );
-    else if (xt > p3 && xt < p4)
+    else if ( xt > p2 && xt < p3 + 1.e-200 )
+        return ( 1. / ( 1. + exp(xt) ) );
+    else if ( xt > p3 && xt < p4 )
         return ( 1. / (2. + xt) );
-    else if (xt > p4 - 1.e-200 && xt < p5)
-        return ( exp(-xt) / (1. + exp(-xt)) );
-    else if (xt > p5 - 1.e-200 && xt < p6)
+    else if ( xt > p4 - 1.e-200 && xt < p5 )
+        return ( exp(-xt) / ( 1. + exp(-xt) ) );
+    else if ( xt > p5 - 1.e-200 && xt < p6 )
         return ( exp(-xt) );
-    else if (xt > p6 - 1.e-200)
+    else if ( xt > p6 - 1.e-200 )
         return 0.;
 }
 
-double swit2_(double wvnmb)
+double swit2_( double wvnmb )
 {
-    return pow( (5.55243*(exp_n(-0.3*wvnmb/15.762) - exp_n(-0.6*wvnmb/15.762)))*wvnmb, 0.5) 
-	  / 1034.66 * pow(sin(wvnmb/65.), 2.);
+    return pow( ( 5.55243 * ( exp_n( -0.3 * wvnmb / 15.762 )
+                - exp_n( -0.6 * wvnmb / 15.762 ) ) ) * wvnmb, 0.5 ) 
+	        / 1034.66 * pow( sin( wvnmb / 65. ), 2. );
 }
 
-
-double GaussianQuad_N2(const double x1, const double x2)
+double GaussianQuad_N2( const double x1, const double x2 )
 {
-    double s=0.0;
+    double s = 0.0;
     double x[6], w[6];
     //gauleg(a2, b2, x, w);
     
     double z1, z, xm, xl, pp, p3, p2, p1;
-    xm=0.5*(x2+x1);
-    xl=0.5*(x2-x1);
-    for(int i=1;i<=3;i++)
+    xm = 0.5 * ( x2 + x1 );
+    xl = 0.5 * ( x2 - x1 );
+    for( int i = 1; i <= 3; i++ )
     {
-		z = (i == 1) ? 0.909632 : -0.0;
-		z = (i == 2) ? 0.540641 : z;
-	    do
+      z = (i == 2) ? 0.540641 : ( (i == 1) ? 0.909632 : -0.0 );
+      do
 	    {
-			p1 = z;
-			p2 = 1;
-			p3 = 0;
-			
-			p3=1;
-			p2=z;
-			p1=((3.0 * z * z) - 1) / 2;
-			
-			p3=p2;
-			p2=p1;
-			p1=((5.0 * z * p2) - (2.0 * z)) / 3;
-			
-			p3=p2;
-			p2=p1;
-			p1=((7.0 * z * p2) - (3.0 * p3)) / 4;
-			
-			p3=p2;
-			p2=p1;
-			p1=((9.0 * z * p2) - (4.0 * p3)) / 5;
-		    
-		    pp=5*(z*p1-p2)/(z*z-1.0);
-		    z1=z;
-		    z=z1-p1/pp;
-	    } while (fabs(z-z1) > 3.0e-11);
+			p1 = ( ( 3.0 * z * z ) - 1 ) / 2;
+			p2 = p1;
+         p1 = ( ( 5.0 * z * p2 ) - ( 2.0 * z ) ) / 3;
+			p3 = p2;
+			p2 = p1;
+			p1 = ( ( 7.0 * z * p2 ) - ( 3.0 * p3 ) ) / 4;
+			p3 = p2;
+			p2 = p1;
+			p1 = ( ( 9.0 * z * p2 ) - ( 4.0 * p3 ) ) / 5;
+         pp = 5 * ( z * p1 - p2 ) / ( z * z - 1.0 );
+		   z1 = z;
+		   z = z1 - p1 / pp;
+	    } while ( fabs( z - z1 ) > 3.0e-11 );
 	    
-	    x[i]=xm-xl*z;
-	    x[5+1-i]=xm+xl*z;
-	    w[i]=2.0*xl/((1.0-z*z)*pp*pp);
-	    w[5+1-i]=w[i];
+	    x[i]       = xm - xl * z;
+	    x[ 5+1-i ] = xm + xl * z;
+	    w[i]       = 2.0 * xl / ( ( 1.0 - z * z ) * pp * pp );
+	    w[ 5+1-i ] = w [i];
     }
     
-    for(int j=1; j<=5; j++) s += w[j]*swit2_(x[j]);
+    for( int j = 1; j <= 5; j++ ) s += w[j] * swit2_( x[j] );
     
     return s;
 }
 
-uint32_t sw2_(int nnounce)
+uint32_t sw2_( int nnounce )
 {
-    double wmax = ((sqrt((double)(nnounce))*(1.+EPSa))/450+100);
-    return ((uint32_t)(GaussianQuad_N2(0., wmax)*(1.+EPSa)*1.e6));
+    double wmax = ( ( sqrt( (double)(nnounce) ) * ( 1.+EPSa ) ) / 450+100 );
+    return ( (uint32_t)( GaussianQuad_N2( 0., wmax ) * ( 1.+EPSa ) * 1.e6 ) );
 }
 
 typedef struct {
@@ -307,8 +296,6 @@ int scanhash_m7m_hash( struct work* work, uint64_t max_nonce,
 
      pdata[19] = n;
 
-// can this be skipped after finding a share? Seems to work ok.
-//out:
      mpf_set_prec_raw(magifpi, prec0);
      mpf_set_prec_raw(magifpi0, prec0);
      mpf_set_prec_raw(mptmp, prec0);
@@ -334,9 +321,8 @@ bool register_m7m_algo( algo_gate_t *gate )
   gate->build_stratum_request = (void*)&std_be_build_stratum_request;
   gate->work_decode           = (void*)&std_be_work_decode;
   gate->submit_getwork_result = (void*)&std_be_submit_getwork_result;
-  gate->set_target            = (void*)&scrypt_set_target;
-  gate->get_max64             = (void*)&get_max64_0x1ffff;
   gate->set_work_data_endian  = (void*)&set_work_data_big_endian;
+  opt_target_factor = 65536.0;
   return true;
 }
 

algo/nist5/zr5.c

@@ -208,12 +208,6 @@ void zr5_get_new_work( struct work* work, struct work* g_work, int thr_id,
        ++(*nonceptr);
 }
 
-int64_t zr5_get_max64 ()
-{
-//  return 0x1ffffLL;
-  return 0x1fffffLL;
-}
-
 void zr5_display_pok( struct work* work )
 {
       if ( work->data[0] & 0x00008000 )
@@ -229,7 +223,6 @@ bool register_zr5_algo( algo_gate_t* gate )
     gate->get_new_work          = (void*)&zr5_get_new_work;
     gate->scanhash              = (void*)&scanhash_zr5;
     gate->hash                  = (void*)&zr5hash;
-    gate->get_max64             = (void*)&zr5_get_max64;
     gate->decode_extra_data     = (void*)&zr5_display_pok;
     gate->build_stratum_request = (void*)&std_be_build_stratum_request;
     gate->work_decode           = (void*)&std_be_work_decode;

algo/quark/anime-4way.c

@@ -49,7 +49,7 @@ void anime_4way_hash( void *state, const void *input )
     __m256i* vhB = (__m256i*)vhashB;
     __m256i vh_mask;
     const uint32_t mask = 8;
-    const __m256i bit3_mask = _mm256_set1_epi64x( 8 );
+    const __m256i bit3_mask = m256_const1_64( 8 );
     const __m256i zero = _mm256_setzero_si256();
     anime_4way_ctx_holder ctx;
     memcpy( &ctx, &anime_4way_ctx, sizeof(anime_4way_ctx) );

algo/quark/hmq1725-4way.c

@@ -21,7 +21,7 @@
 #include "algo/shabal/shabal-hash-4way.h"
 #include "algo/whirlpool/sph_whirlpool.h"
 #include "algo/haval/haval-hash-4way.h"
-#include "algo/sha/sha2-hash-4way.h"
+#include "algo/sha/sha-hash-4way.h"
 
 union _hmq1725_4way_context_overlay
 {
@@ -57,7 +57,7 @@ extern void hmq1725_4way_hash(void *state, const void *input)
      uint32_t vhashB[32<<2] __attribute__ ((aligned (64)));
      hmq1725_4way_context_overlay ctx __attribute__ ((aligned (64)));
      __m256i vh_mask;     
-     const __m256i vmask = _mm256_set1_epi64x( 24 );
+     const __m256i vmask = m256_const1_64( 24 );
      const uint32_t mask = 24;
      __m256i* vh  = (__m256i*)vhash;
      __m256i* vhA = (__m256i*)vhashA;

algo/quark/hmq1725-gate.c

@@ -10,8 +10,8 @@ bool register_hmq1725_algo( algo_gate_t* gate )
   gate->scanhash  = (void*)&scanhash_hmq1725;
   gate->hash      = (void*)&hmq1725hash;
 #endif
-  gate->set_target       = (void*)&scrypt_set_target;
   gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT;
+  opt_target_factor = 65536.0;
   return true;
 };
 

algo/quark/hmq1725.c

@@ -409,14 +409,3 @@ int scanhash_hmq1725( struct work *work, uint32_t max_nonce,
 	pdata[19] = n;
 	return 0;
 }
-/*
-bool register_hmq1725_algo( algo_gate_t* gate )
-{
-  init_hmq1725_ctx();
-  gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT;
-  gate->set_target       = (void*)&scrypt_set_target;
-  gate->scanhash         = (void*)&scanhash_hmq1725;
-  gate->hash             = (void*)&hmq1725hash;
-  return true;
-};
-*/

algo/quark/quark-4way.c

@@ -49,7 +49,7 @@ void quark_4way_hash( void *state, const void *input )
     __m256i* vhB = (__m256i*)vhashB;
     __m256i vh_mask;
     quark_4way_ctx_holder ctx;
-    const __m256i bit3_mask = _mm256_set1_epi64x( 8 );
+    const __m256i bit3_mask = m256_const1_64( 8 );
     const uint32_t mask = 8;
     const __m256i zero = _mm256_setzero_si256();
 

algo/ripemd/lbry-4way.c

@@ -3,7 +3,7 @@
 #include <stdint.h>
 #include <string.h>
 #include <stdio.h>
-#include "algo/sha/sha2-hash-4way.h"
+#include "algo/sha/sha-hash-4way.h"
 #include "ripemd-hash-4way.h"
 
 #define LBRY_INPUT_SIZE 112

algo/ripemd/lbry-gate.c

@@ -41,6 +41,7 @@ void lbry_le_build_stratum_request( char *req, struct work *work,
    free(xnonce2str);
 }
 
+/*
 void lbry_build_block_header( struct work* g_work, uint32_t version,
                              uint32_t *prevhash, uint32_t *merkle_root,
                              uint32_t ntime, uint32_t nbits )
@@ -63,6 +64,7 @@ void lbry_build_block_header( struct work* g_work, uint32_t version,
    g_work->data[ LBRY_NBITS_INDEX ] = nbits;
    g_work->data[28] = 0x80000000;
 }
+*/
 
 void lbry_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
 {
@@ -92,13 +94,6 @@ void lbry_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
    g_work->data[28] = 0x80000000;
 }
 
-void lbry_set_target( struct work* work, double job_diff )
-{
- work_set_target( work, job_diff / (256.0 * opt_diff_factor) );
-}
-
-int64_t lbry_get_max64() { return 0x1ffffLL; }
-
 int lbry_get_work_data_size() { return LBRY_WORK_DATA_SIZE; }
 
 bool register_lbry_algo( algo_gate_t* gate )
@@ -115,15 +110,14 @@ bool register_lbry_algo( algo_gate_t* gate )
   gate->hash                  = (void*)&lbry_hash;
 #endif
   gate->calc_network_diff     = (void*)&lbry_calc_network_diff;
-  gate->get_max64             = (void*)&lbry_get_max64;
   gate->build_stratum_request = (void*)&lbry_le_build_stratum_request;
 //  gate->build_block_header    = (void*)&build_block_header;
   gate->build_extraheader     = (void*)&lbry_build_extraheader;
-  gate->set_target            = (void*)&lbry_set_target;
   gate->ntime_index           = LBRY_NTIME_INDEX;
   gate->nbits_index           = LBRY_NBITS_INDEX;
   gate->nonce_index           = LBRY_NONCE_INDEX;
   gate->get_work_data_size    = (void*)&lbry_get_work_data_size;
+  opt_target_factor = 256.0;
   return true;
 }
 

algo/ripemd/ripemd-hash-4way.c

@@ -5,23 +5,26 @@
 #include <stddef.h>
 #include <string.h>
 
+/*
 static const uint32_t IV[5] =
 { 0x67452301, 0xEFCDAB89, 0x98BADCFE, 0x10325476, 0xC3D2E1F0 };
+*/
 
 /*
  * Round constants for RIPEMD-160.
  */
-#define K11  0x00000000
-#define K12  0x5A827999
-#define K13  0x6ED9EBA1
-#define K14  0x8F1BBCDC
-#define K15  0xA953FD4E
 
-#define K21  0x50A28BE6
-#define K22  0x5C4DD124
-#define K23  0x6D703EF3
-#define K24  0x7A6D76E9
-#define K25  0x00000000
+#define K11  0x0000000000000000
+#define K12  0x5A8279995A827999
+#define K13  0x6ED9EBA16ED9EBA1
+#define K14  0x8F1BBCDC8F1BBCDC
+#define K15  0xA953FD4EA953FD4E
+
+#define K21  0x50A28BE650A28BE6
+#define K22  0x5C4DD1245C4DD124
+#define K23  0x6D703EF36D703EF3
+#define K24  0x7A6D76E97A6D76E9
+#define K25  0x0000000000000000
 
 // RIPEMD-160 4 way
 
@@ -44,7 +47,7 @@ static const uint32_t IV[5] =
 do{ \
    a = _mm_add_epi32( mm128_rol_32( _mm_add_epi32( _mm_add_epi32( \
                 _mm_add_epi32( a, f( b ,c, d ) ), r ), \
-                                 _mm_set1_epi32( k ) ), s ), e ); \
+                                 m128_const1_64( k ) ), s ), e ); \
    c = mm128_rol_32( c, 10 );\
 } while (0)
 
@@ -248,11 +251,11 @@ static void ripemd160_4way_round( ripemd160_4way_context *sc )
 
 void ripemd160_4way_init( ripemd160_4way_context *sc )
 {
-   sc->val[0] = _mm_set1_epi32( IV[0] );
-   sc->val[1] = _mm_set1_epi32( IV[1] );
-   sc->val[2] = _mm_set1_epi32( IV[2] );
-   sc->val[3] = _mm_set1_epi32( IV[3] );
-   sc->val[4] = _mm_set1_epi32( IV[4] );
+   sc->val[0] = m128_const1_64( 0x6745230167452301 );
+   sc->val[1] = m128_const1_64( 0xEFCDAB89EFCDAB89 );
+   sc->val[2] = m128_const1_64( 0x98BADCFE98BADCFE );
+   sc->val[3] = m128_const1_64( 0x1032547610325476 );
+   sc->val[4] = m128_const1_64( 0xC3D2E1F0C3D2E1F0 );
    sc->count_high = sc->count_low = 0;
 }
 
@@ -343,7 +346,7 @@ void ripemd160_4way_close( ripemd160_4way_context  *sc, void *dst )
 do{ \
    a = _mm256_add_epi32( mm256_rol_32( _mm256_add_epi32( _mm256_add_epi32( \
                 _mm256_add_epi32( a, f( b ,c, d ) ), r ), \
-                                 _mm256_set1_epi32( k ) ), s ), e ); \
+                                 m256_const1_64( k ) ), s ), e ); \
    c = mm256_rol_32( c, 10 );\
 } while (0)
     
@@ -548,11 +551,11 @@ static void ripemd160_8way_round( ripemd160_8way_context *sc )
 
 void ripemd160_8way_init( ripemd160_8way_context *sc )
 {
-   sc->val[0] = _mm256_set1_epi32( IV[0] );
-   sc->val[1] = _mm256_set1_epi32( IV[1] );
-   sc->val[2] = _mm256_set1_epi32( IV[2] );
-   sc->val[3] = _mm256_set1_epi32( IV[3] );
-   sc->val[4] = _mm256_set1_epi32( IV[4] );
+   sc->val[0] = m256_const1_64( 0x6745230167452301 );
+   sc->val[1] = m256_const1_64( 0xEFCDAB89EFCDAB89 );
+   sc->val[2] = m256_const1_64( 0x98BADCFE98BADCFE );
+   sc->val[3] = m256_const1_64( 0x1032547610325476 );
+   sc->val[4] = m256_const1_64( 0xC3D2E1F0C3D2E1F0 );
    sc->count_high = sc->count_low = 0;
 }
 

algo/scrypt/neoscrypt.c

@@ -1070,17 +1070,6 @@ int scanhash_neoscrypt( struct work *work,
     return 0;
 }
 
-int64_t get_neoscrypt_max64() { return 0x3ffff; }
-
-void neoscrypt_wait_for_diff( struct stratum_ctx *stratum )
-{
-   while ( !stratum->job.diff )
-   {
-//      applog(LOG_DEBUG, "Waiting for Stratum to set the job difficulty");
-      sleep(1);
-   }
-}
-
 int neoscrypt_get_work_data_size () { return 80; }
 
 bool register_neoscrypt_algo( algo_gate_t* gate )
@@ -1088,14 +1077,12 @@ bool register_neoscrypt_algo( algo_gate_t* gate )
   gate->optimizations         = SSE2_OPT;
   gate->scanhash              = (void*)&scanhash_neoscrypt;
   gate->hash                  = (void*)&neoscrypt;
-  gate->get_max64             = (void*)&get_neoscrypt_max64;
-  gate->set_target            = (void*)&scrypt_set_target;
-  gate->wait_for_diff         = (void*)&neoscrypt_wait_for_diff;
   gate->build_stratum_request = (void*)&std_be_build_stratum_request;
   gate->work_decode           = (void*)&std_be_work_decode;
   gate->submit_getwork_result = (void*)&std_be_submit_getwork_result;
   gate->set_work_data_endian  = (void*)&set_work_data_big_endian;
   gate->get_work_data_size    = (void*)&neoscrypt_get_work_data_size;
+  opt_target_factor = 65536.0;
   return true;
 };
 

algo/scrypt/pluck.c

@@ -483,11 +483,6 @@ int scanhash_pluck( struct work *work, uint32_t max_nonce,
 	return 0;
 }
 
-int64_t pluck_get_max64 ()
-{
-  return 0x1ffLL;
-}
-
 bool pluck_miner_thread_init( int thr_id )
 { 
   scratchbuf = malloc( 128 * 1024 ); 
@@ -503,8 +498,7 @@ bool register_pluck_algo( algo_gate_t* gate )
   gate->miner_thread_init = (void*)&pluck_miner_thread_init;
   gate->scanhash         = (void*)&scanhash_pluck;
   gate->hash             = (void*)&pluck_hash;
-  gate->set_target       = (void*)&scrypt_set_target;
-  gate->get_max64        = (void*)&pluck_get_max64;
+  opt_target_factor = 65536.0;
   return true;
 };
 

algo/scrypt/scrypt.c

@@ -698,8 +698,8 @@ static void scrypt_1024_1_1_256_24way(const uint32_t *input,
 extern int scanhash_scrypt( struct work *work, uint32_t max_nonce,
                             uint64_t *hashes_done, struct thr_info *mythr )
 {
-        uint32_t *pdata = work->data;
-        uint32_t *ptarget = work->target;
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
 	uint32_t data[SCRYPT_MAX_WAYS * 20], hash[SCRYPT_MAX_WAYS * 8];
 	uint32_t midstate[8];
 	uint32_t n = pdata[19] - 1;
@@ -766,8 +766,6 @@ extern int scanhash_scrypt( struct work *work, uint32_t max_nonce,
 	return 0;
 }
 
-int64_t scrypt_get_max64() { return 0xfff; }
-
 bool scrypt_miner_thread_init( int thr_id )
 {
  scratchbuf = scrypt_buffer_alloc( scratchbuf_size );  
@@ -783,13 +781,16 @@ bool register_scrypt_algo( algo_gate_t* gate )
   gate->miner_thread_init =(void*)&scrypt_miner_thread_init;
   gate->scanhash         = (void*)&scanhash_scrypt;
 //  gate->hash             = (void*)&scrypt_1024_1_1_256_24way;
-  gate->set_target       = (void*)&scrypt_set_target;
-  gate->get_max64        = (void*)&scrypt_get_max64;
+  opt_target_factor = 65536.0;
 
-  if ( !opt_scrypt_n )
+  if ( !opt_param_n )
+  {
+     opt_param_n = 1024;
      scratchbuf_size = 1024;
+  }
   else
-     scratchbuf_size = opt_scrypt_n;
+     scratchbuf_size = opt_param_n;
+  applog(LOG_INFO,"Scrypt paramaters: N= %d, R= 1.", opt_param_n );
   return true;
 };
 

algo/scryptjane/scrypt-jane-chacha.h

@@ -55,6 +55,7 @@ typedef uint32_t scrypt_mix_word_t;
 	#include "scrypt-jane-romix-template.h"
 #endif
 
+
 /* cpu agnostic */
 #define SCRYPT_ROMIX_FN scrypt_ROMix_basic
 #define SCRYPT_MIX_FN chacha_core_basic

algo/scryptjane/scrypt-jane-romix-template.h

@@ -1,9 +1,11 @@
 #if !defined(SCRYPT_CHOOSE_COMPILETIME) || !defined(SCRYPT_HAVE_ROMIX)
 
+/*
 #if defined(SCRYPT_CHOOSE_COMPILETIME)
 #undef SCRYPT_ROMIX_FN
 #define SCRYPT_ROMIX_FN scrypt_ROMix
 #endif
+*/
 
 #undef SCRYPT_HAVE_ROMIX
 #define SCRYPT_HAVE_ROMIX

algo/scryptjane/scrypt-jane.c

@@ -240,24 +240,23 @@ bool register_scryptjane_algo( algo_gate_t* gate )
 {
     gate->scanhash   = (void*)&scanhash_scryptjane;
     gate->hash       = (void*)&scryptjanehash;
-    gate->set_target = (void*)&scrypt_set_target;
-    gate->get_max64  = (void*)&get_max64_0x40LL;
+    opt_target_factor = 65536.0;
 
     // figure out if arg in N or Nfactor
-    if ( !opt_scrypt_n )
+    if ( !opt_param_n )
     {
       applog( LOG_ERR, "The N factor must be specified in the form algo:nf");
       return false;
     }
-    else if ( opt_scrypt_n < 32 )
+    else if ( opt_param_n < 32 )
     {
       // arg is Nfactor, calculate N
-      sj_N = 1 << ( opt_scrypt_n + 1 );
+      sj_N = 1 << ( opt_param_n + 1 );
     }
     else
     {
       // arg is N
-      sj_N = opt_scrypt_n;
+      sj_N = opt_param_n;
     }
     return true;
 }

algo/sha/sha-hash-4way.h

@@ -55,32 +55,13 @@ typedef struct {
    __m128i buf[64>>2];
    __m128i val[8];
    uint32_t count_high, count_low;
+   bool initialized;
 } sha256_4way_context;
 
 void sha256_4way_init( sha256_4way_context *sc );
 void sha256_4way( sha256_4way_context *sc, const void *data, size_t len );
 void sha256_4way_close( sha256_4way_context *sc, void *dst );
 
-/*
-// SHA-256 7 way hybrid
-// Combines SSE, MMX and scalar data to do 8 + 2 + 1 parallel.
-typedef struct {
-   __m128i  bufx[64>>2];
-   __m128i  valx[8];
-   __m64    bufy[64>>2];
-   __m64    valy[8];
-   uint32_t bufz[64>>2];
-   uint32_t valz[8];
-   uint32_t count_high, count_low;
-} sha256_7way_context;
-
-void sha256_7way_init( sha256_7way_context *ctx );
-void sha256_7way( sha256_7way_context *ctx, const void *datax,
-                         void *datay, void *dataz, size_t len );
-void sha256_7way_close( sha256_7way_context *ctx, void *dstx, void *dstyx,
-                         void *dstz  );
-*/
-
 #if defined (__AVX2__)
 
 // SHA-256 8 way
@@ -89,6 +70,7 @@ typedef struct {
    __m256i buf[64>>2];
    __m256i val[8];
    uint32_t count_high, count_low;
+   bool initialized;
 } sha256_8way_context;
 
 void sha256_8way_init( sha256_8way_context *sc );
@@ -103,6 +85,7 @@ typedef struct {
    __m256i buf[128>>3];
    __m256i val[8];
    uint64_t count;
+   bool initialized;
 } sha512_4way_context;
 
 void sha512_4way_init( sha512_4way_context *sc);

algo/sha/sha2.c

@@ -12,6 +12,7 @@
 
 #include <string.h>
 #include <inttypes.h>
+#include <openssl/sha.h>
 
 #if defined(USE_ASM) && defined(__arm__) && defined(__APCS_32__)
 #define EXTERN_SHA256
@@ -197,7 +198,17 @@ static void sha256d_80_swap(uint32_t *hash, const uint32_t *data)
 
 extern void sha256d(unsigned char *hash, const unsigned char *data, int len)
 {
-	uint32_t S[16], T[16];
+#if defined(__SHA__)
+   SHA256_CTX ctx;
+   SHA256_Init( &ctx );
+   SHA256_Update( &ctx, data, len );
+   SHA256_Final( (unsigned char*)hash, &ctx );
+   SHA256_Init( &ctx );
+   SHA256_Update( &ctx, hash, 32 );
+   SHA256_Final( (unsigned char*)hash, &ctx );
+#else
+
+   uint32_t S[16], T[16];
 	int i, r;
 
 	sha256_init(S);
@@ -218,6 +229,7 @@ extern void sha256d(unsigned char *hash, const unsigned char *data, int len)
 	sha256_transform(T, S, 0);
 	for (i = 0; i < 8; i++)
 		be32enc((uint32_t *)hash + i, T[i]);
+#endif
 }
 
 static inline void sha256d_preextend(uint32_t *W)
@@ -635,9 +647,46 @@ int scanhash_sha256d( struct work *work,
 	return 0;
 }
 
+int scanhash_SHA256d( struct work *work, const uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t _ALIGN(128) hash[8];
+   uint32_t _ALIGN(64) data[20];
+   uint32_t *pdata = work->data;
+   const uint32_t *ptarget = work->target;
+   uint32_t n = pdata[19] - 1;
+   const uint32_t first_nonce = pdata[19];
+   const uint32_t Htarg = ptarget[7];
+   int thr_id = mythr->id;  // thr_id arg is deprecated
+
+   memcpy( data, pdata, 80 );
+
+   do {
+      data[19] = ++n;
+      sha256d( (unsigned char*)hash, (const unsigned char*)data, 80 );
+      if ( unlikely( swab32( hash[7] ) <= Htarg ) )
+      {
+         pdata[19] = n;
+         sha256d_80_swap(hash, pdata);
+         if ( fulltest( hash, ptarget ) && !opt_benchmark )
+            submit_solution( work, hash, mythr );
+      }
+   } while ( likely( n < max_nonce && !work_restart[thr_id].restart ) );
+   *hashes_done = n - first_nonce + 1;
+   pdata[19] = n;
+   return 0;
+}
+
+
 bool register_sha256d_algo( algo_gate_t* gate )
 {
-    gate->scanhash = (void*)&scanhash_sha256d;
+#if defined(__SHA__)
+   gate->optimizations = SHA_OPT;
+   gate->scanhash = (void*)&scanhash_SHA256d;
+#else
+   gate->optimizations = SSE2_OPT | AVX2_OPT;
+   gate->scanhash = (void*)&scanhash_sha256d;
+#endif
     gate->hash     = (void*)&sha256d;
     return true;
 };

algo/sha/sha256-hash-4way.c

@@ -34,19 +34,18 @@
 
 #include <stddef.h>
 #include <string.h>
-
-#include "sha2-hash-4way.h"
-
-#include <stdio.h>
+#include "sha-hash-4way.h"
 
 // SHA-256 32 bit
 
+/*
 static const sph_u32 H256[8] = {
         SPH_C32(0x6A09E667), SPH_C32(0xBB67AE85),
         SPH_C32(0x3C6EF372), SPH_C32(0xA54FF53A),
         SPH_C32(0x510E527F), SPH_C32(0x9B05688C),
         SPH_C32(0x1F83D9AB), SPH_C32(0x5BE0CD19)
 };
+*/
 
 static const sph_u32 K256[64] = {
         SPH_C32(0x428A2F98), SPH_C32(0x71374491),
@@ -113,16 +112,17 @@ static const sph_u32 K256[64] = {
 
 #define SHA2s_4WAY_STEP(A, B, C, D, E, F, G, H, i, j) \
 do { \
-  register __m128i T1, T2; \
+  __m128i T1, T2; \
+  __m128i K = _mm_set1_epi32( K256[( (j)+(i) )] ); \
   T1 = _mm_add_epi32( H, mm128_add4_32( BSG2_1(E), CHs(E, F, G), \
-                             _mm_set1_epi32( K256[( (j)+(i) )] ), W[i] ) ); \
+                                        K, W[i] ) ); \
   T2 = _mm_add_epi32( BSG2_0(A), MAJs(A, B, C) ); \
   D  = _mm_add_epi32( D,  T1 ); \
   H  = _mm_add_epi32( T1, T2 ); \
 } while (0)
 
 static void
-sha256_4way_round( __m128i *in, __m128i r[8] )
+sha256_4way_round( sha256_4way_context *ctx, __m128i *in, __m128i r[8] )
 {
    register  __m128i A, B, C, D, E, F, G, H;
    __m128i W[16];
@@ -130,14 +130,28 @@ sha256_4way_round( __m128i *in, __m128i r[8] )
    mm128_block_bswap_32( W, in );
    mm128_block_bswap_32( W+8, in+8 );
 
-   A = r[0];
-   B = r[1];
-   C = r[2];
-   D = r[3];
-   E = r[4];
-   F = r[5];
-   G = r[6];
-   H = r[7];
+   if ( ctx->initialized )
+   {
+      A = r[0];
+      B = r[1];
+      C = r[2];
+      D = r[3];
+      E = r[4];
+      F = r[5];
+      G = r[6];
+      H = r[7];
+   }
+   else
+   {
+      A = m128_const1_64( 0x6A09E6676A09E667 );
+      B = m128_const1_64( 0xBB67AE85BB67AE85 );
+      C = m128_const1_64( 0x3C6EF3723C6EF372 );
+      D = m128_const1_64( 0xA54FF53AA54FF53A );
+      E = m128_const1_64( 0x510E527F510E527F );
+      F = m128_const1_64( 0x9B05688C9B05688C );
+      G = m128_const1_64( 0x1F83D9AB1F83D9AB );
+      H = m128_const1_64( 0x5BE0CD195BE0CD19 );
+   }
 
    SHA2s_4WAY_STEP( A, B, C, D, E, F, G, H,  0, 0 );
    SHA2s_4WAY_STEP( H, A, B, C, D, E, F, G,  1, 0 );
@@ -193,19 +207,36 @@ sha256_4way_round( __m128i *in, __m128i r[8] )
       SHA2s_4WAY_STEP( B, C, D, E, F, G, H, A, 15, j );
    }
 
-   r[0] = _mm_add_epi32( r[0], A );
-   r[1] = _mm_add_epi32( r[1], B );
-   r[2] = _mm_add_epi32( r[2], C );
-   r[3] = _mm_add_epi32( r[3], D );
-   r[4] = _mm_add_epi32( r[4], E );
-   r[5] = _mm_add_epi32( r[5], F );
-   r[6] = _mm_add_epi32( r[6], G );
-   r[7] = _mm_add_epi32( r[7], H );
+   if ( ctx->initialized )
+   {
+      r[0] = _mm_add_epi32( r[0], A );
+      r[1] = _mm_add_epi32( r[1], B );
+      r[2] = _mm_add_epi32( r[2], C );
+      r[3] = _mm_add_epi32( r[3], D );
+      r[4] = _mm_add_epi32( r[4], E );
+      r[5] = _mm_add_epi32( r[5], F );
+      r[6] = _mm_add_epi32( r[6], G );
+      r[7] = _mm_add_epi32( r[7], H );
+   }
+   else
+   {
+      ctx->initialized = true;
+      r[0] = _mm_add_epi32( A, m128_const1_64( 0x6A09E6676A09E667 ) );
+      r[1] = _mm_add_epi32( B, m128_const1_64( 0xBB67AE85BB67AE85 ) );
+      r[2] = _mm_add_epi32( C, m128_const1_64( 0x3C6EF3723C6EF372 ) );
+      r[3] = _mm_add_epi32( D, m128_const1_64( 0xA54FF53AA54FF53A ) );
+      r[4] = _mm_add_epi32( E, m128_const1_64( 0x510E527F510E527F ) );
+      r[5] = _mm_add_epi32( F, m128_const1_64( 0x9B05688C9B05688C ) );
+      r[6] = _mm_add_epi32( G, m128_const1_64( 0x1F83D9AB1F83D9AB ) );
+      r[7] = _mm_add_epi32( H, m128_const1_64( 0x5BE0CD195BE0CD19 ) );
+   }
 }
 
 void sha256_4way_init( sha256_4way_context *sc )
 {
+   sc->initialized = false;
    sc->count_high = sc->count_low = 0;
+/*
    sc->val[0] = _mm_set1_epi32( H256[0] );
    sc->val[1] = _mm_set1_epi32( H256[1] );
    sc->val[2] = _mm_set1_epi32( H256[2] );
@@ -214,6 +245,7 @@ void sha256_4way_init( sha256_4way_context *sc )
    sc->val[5] = _mm_set1_epi32( H256[5] );
    sc->val[6] = _mm_set1_epi32( H256[6] );
    sc->val[7] = _mm_set1_epi32( H256[7] );
+*/
 }
 
 void sha256_4way( sha256_4way_context *sc, const void *data, size_t len )
@@ -237,7 +269,7 @@ void sha256_4way( sha256_4way_context *sc, const void *data, size_t len )
       len -= clen;
       if ( ptr == buf_size )
       {
-         sha256_4way_round( sc->buf, sc->val );
+         sha256_4way_round( sc, sc->buf, sc->val );
          ptr = 0;
       }
       clow = sc->count_low;
@@ -256,13 +288,13 @@ void sha256_4way_close( sha256_4way_context *sc, void *dst )
     const int pad = buf_size - 8;
 
     ptr = (unsigned)sc->count_low & (buf_size - 1U);
-    sc->buf[ ptr>>2 ] = _mm_set1_epi32( 0x80 );
+    sc->buf[ ptr>>2 ] = m128_const1_64( 0x0000008000000080 );
     ptr += 4;
 
     if ( ptr > pad )
     {
          memset_zero_128( sc->buf + (ptr>>2), (buf_size - ptr) >> 2 );
-         sha256_4way_round( sc->buf, sc->val );
+         sha256_4way_round( sc, sc->buf, sc->val );
          memset_zero_128( sc->buf, pad >> 2 );
     }
     else
@@ -273,10 +305,12 @@ void sha256_4way_close( sha256_4way_context *sc, void *dst )
     low = low << 3;
 
     sc->buf[ pad >> 2 ] =
-                 mm128_bswap_32( _mm_set1_epi32( high ) );
+                 mm128_bswap_32( m128_const1_32( high ) );
+//                 mm128_bswap_32( _mm_set1_epi32( high ) );
     sc->buf[ ( pad+4 ) >> 2 ] =
-                 mm128_bswap_32( _mm_set1_epi32( low ) );
-    sha256_4way_round( sc->buf, sc->val );
+                 mm128_bswap_32( m128_const1_32( low ) );
+//                 mm128_bswap_32( _mm_set1_epi32( low ) );
+    sha256_4way_round( sc, sc->buf, sc->val );
 
     mm128_block_bswap_32( dst, sc->val );
 }
@@ -313,16 +347,17 @@ void sha256_4way_close( sha256_4way_context *sc, void *dst )
 
 #define SHA2s_8WAY_STEP(A, B, C, D, E, F, G, H, i, j) \
 do { \
-  register __m256i T1, T2; \
-   T1 = _mm256_add_epi32( H, mm256_add4_32( BSG2_1x(E), CHx(E, F, G), \
-                          _mm256_set1_epi32( K256[( (j)+(i) )] ), W[i] ) ); \
+  __m256i T1, T2; \
+  __m256i K = _mm256_set1_epi32( K256[( (j)+(i) )] ); \
+  T1 = _mm256_add_epi32( H, mm256_add4_32( BSG2_1x(E), CHx(E, F, G), \
+                                           K, W[i] ) ); \
   T2 = _mm256_add_epi32( BSG2_0x(A), MAJx(A, B, C) ); \
   D  = _mm256_add_epi32( D,  T1 ); \
   H  = _mm256_add_epi32( T1, T2 ); \
 } while (0)
 
 static void
-sha256_8way_round( __m256i *in, __m256i r[8] )
+sha256_8way_round( sha256_8way_context *ctx,  __m256i *in, __m256i r[8] )
 {
    register  __m256i A, B, C, D, E, F, G, H;
    __m256i W[16];
@@ -330,14 +365,28 @@ sha256_8way_round( __m256i *in, __m256i r[8] )
    mm256_block_bswap_32( W  , in   );
    mm256_block_bswap_32( W+8, in+8 );
 
-   A = r[0];
-   B = r[1];
-   C = r[2];
-   D = r[3];
-   E = r[4];
-   F = r[5];
-   G = r[6];
-   H = r[7];
+   if ( ctx->initialized )
+   {
+      A = r[0];
+      B = r[1];
+      C = r[2];
+      D = r[3];
+      E = r[4];
+      F = r[5];
+      G = r[6];
+      H = r[7];
+   }
+   else
+   {
+      A = m256_const1_64( 0x6A09E6676A09E667 );
+      B = m256_const1_64( 0xBB67AE85BB67AE85 );
+      C = m256_const1_64( 0x3C6EF3723C6EF372 );
+      D = m256_const1_64( 0xA54FF53AA54FF53A );
+      E = m256_const1_64( 0x510E527F510E527F );
+      F = m256_const1_64( 0x9B05688C9B05688C );
+      G = m256_const1_64( 0x1F83D9AB1F83D9AB );
+      H = m256_const1_64( 0x5BE0CD195BE0CD19 );
+   }
 
    SHA2s_8WAY_STEP( A, B, C, D, E, F, G, H,  0, 0 );
    SHA2s_8WAY_STEP( H, A, B, C, D, E, F, G,  1, 0 );
@@ -393,20 +442,36 @@ sha256_8way_round( __m256i *in, __m256i r[8] )
       SHA2s_8WAY_STEP( B, C, D, E, F, G, H, A, 15, j );
    }
 
-   r[0] = _mm256_add_epi32( r[0], A );
-   r[1] = _mm256_add_epi32( r[1], B );
-   r[2] = _mm256_add_epi32( r[2], C );
-   r[3] = _mm256_add_epi32( r[3], D );
-   r[4] = _mm256_add_epi32( r[4], E );
-   r[5] = _mm256_add_epi32( r[5], F );
-   r[6] = _mm256_add_epi32( r[6], G );
-   r[7] = _mm256_add_epi32( r[7], H );
+   if ( ctx->initialized )
+   {
+      r[0] = _mm256_add_epi32( r[0], A );
+      r[1] = _mm256_add_epi32( r[1], B );
+      r[2] = _mm256_add_epi32( r[2], C );
+      r[3] = _mm256_add_epi32( r[3], D );
+      r[4] = _mm256_add_epi32( r[4], E );
+      r[5] = _mm256_add_epi32( r[5], F );
+      r[6] = _mm256_add_epi32( r[6], G );
+      r[7] = _mm256_add_epi32( r[7], H );
+   }
+   else
+   {
+      ctx->initialized = true;
+      r[0] = _mm256_add_epi32( A, m256_const1_64( 0x6A09E6676A09E667 ) );
+      r[1] = _mm256_add_epi32( B, m256_const1_64( 0xBB67AE85BB67AE85 ) );
+      r[2] = _mm256_add_epi32( C, m256_const1_64( 0x3C6EF3723C6EF372 ) );
+      r[3] = _mm256_add_epi32( D, m256_const1_64( 0xA54FF53AA54FF53A ) );
+      r[4] = _mm256_add_epi32( E, m256_const1_64( 0x510E527F510E527F ) );
+      r[5] = _mm256_add_epi32( F, m256_const1_64( 0x9B05688C9B05688C ) );
+      r[6] = _mm256_add_epi32( G, m256_const1_64( 0x1F83D9AB1F83D9AB ) );
+      r[7] = _mm256_add_epi32( H, m256_const1_64( 0x5BE0CD195BE0CD19 ) );
+   }
 }
 
-
 void sha256_8way_init( sha256_8way_context *sc )
 {
+   sc->initialized = false;
    sc->count_high = sc->count_low = 0;
+/*
    sc->val[0] = _mm256_set1_epi32( H256[0] );
    sc->val[1] = _mm256_set1_epi32( H256[1] );
    sc->val[2] = _mm256_set1_epi32( H256[2] );
@@ -415,6 +480,7 @@ void sha256_8way_init( sha256_8way_context *sc )
    sc->val[5] = _mm256_set1_epi32( H256[5] );
    sc->val[6] = _mm256_set1_epi32( H256[6] );
    sc->val[7] = _mm256_set1_epi32( H256[7] );
+*/
 }
 
 void sha256_8way( sha256_8way_context *sc, const void *data, size_t len )
@@ -438,7 +504,7 @@ void sha256_8way( sha256_8way_context *sc, const void *data, size_t len )
       len -= clen;
       if ( ptr == buf_size )
       {
-         sha256_8way_round( sc->buf, sc->val );
+         sha256_8way_round( sc, sc->buf, sc->val );
          ptr = 0;
       }
       clow = sc->count_low;
@@ -457,13 +523,13 @@ void sha256_8way_close( sha256_8way_context *sc, void *dst )
     const int pad = buf_size - 8;
 
     ptr = (unsigned)sc->count_low & (buf_size - 1U);
-    sc->buf[ ptr>>2 ] = _mm256_set1_epi32( 0x80 );
+    sc->buf[ ptr>>2 ] = m256_const1_64( 0x0000008000000080 );
     ptr += 4;
 
     if ( ptr > pad )
     {
          memset_zero_256( sc->buf + (ptr>>2), (buf_size - ptr) >> 2 );
-         sha256_8way_round( sc->buf, sc->val );
+         sha256_8way_round( sc, sc->buf, sc->val );
          memset_zero_256( sc->buf, pad >> 2 );
     }
     else
@@ -474,211 +540,14 @@ void sha256_8way_close( sha256_8way_context *sc, void *dst )
     low = low << 3;
 
     sc->buf[ pad >> 2 ] =
-                 mm256_bswap_32( _mm256_set1_epi32( high ) );
+                 mm256_bswap_32( m256_const1_32( high ) );
     sc->buf[ ( pad+4 ) >> 2 ] =
-                 mm256_bswap_32( _mm256_set1_epi32( low ) );
+                 mm256_bswap_32( m256_const1_32( low ) );
 
-    sha256_8way_round( sc->buf, sc->val );
+    sha256_8way_round( sc, sc->buf, sc->val );
 
     mm256_block_bswap_32( dst, sc->val );
 }
 
-
-// SHA-512 4 way 64 bit
-
-static const sph_u64 H512[8] = {
-        SPH_C64(0x6A09E667F3BCC908), SPH_C64(0xBB67AE8584CAA73B),
-        SPH_C64(0x3C6EF372FE94F82B), SPH_C64(0xA54FF53A5F1D36F1),
-        SPH_C64(0x510E527FADE682D1), SPH_C64(0x9B05688C2B3E6C1F),
-        SPH_C64(0x1F83D9ABFB41BD6B), SPH_C64(0x5BE0CD19137E2179)
-};
-
-static const sph_u64 K512[80] = {
-	SPH_C64(0x428A2F98D728AE22), SPH_C64(0x7137449123EF65CD),
-	SPH_C64(0xB5C0FBCFEC4D3B2F), SPH_C64(0xE9B5DBA58189DBBC),
-	SPH_C64(0x3956C25BF348B538), SPH_C64(0x59F111F1B605D019),
-	SPH_C64(0x923F82A4AF194F9B), SPH_C64(0xAB1C5ED5DA6D8118),
-	SPH_C64(0xD807AA98A3030242), SPH_C64(0x12835B0145706FBE),
-	SPH_C64(0x243185BE4EE4B28C), SPH_C64(0x550C7DC3D5FFB4E2),
-	SPH_C64(0x72BE5D74F27B896F), SPH_C64(0x80DEB1FE3B1696B1),
-	SPH_C64(0x9BDC06A725C71235), SPH_C64(0xC19BF174CF692694),
-	SPH_C64(0xE49B69C19EF14AD2), SPH_C64(0xEFBE4786384F25E3),
-	SPH_C64(0x0FC19DC68B8CD5B5), SPH_C64(0x240CA1CC77AC9C65),
-	SPH_C64(0x2DE92C6F592B0275), SPH_C64(0x4A7484AA6EA6E483),
-	SPH_C64(0x5CB0A9DCBD41FBD4), SPH_C64(0x76F988DA831153B5),
-	SPH_C64(0x983E5152EE66DFAB), SPH_C64(0xA831C66D2DB43210),
-	SPH_C64(0xB00327C898FB213F), SPH_C64(0xBF597FC7BEEF0EE4),
-	SPH_C64(0xC6E00BF33DA88FC2), SPH_C64(0xD5A79147930AA725),
-	SPH_C64(0x06CA6351E003826F), SPH_C64(0x142929670A0E6E70),
-	SPH_C64(0x27B70A8546D22FFC), SPH_C64(0x2E1B21385C26C926),
-	SPH_C64(0x4D2C6DFC5AC42AED), SPH_C64(0x53380D139D95B3DF),
-	SPH_C64(0x650A73548BAF63DE), SPH_C64(0x766A0ABB3C77B2A8),
-	SPH_C64(0x81C2C92E47EDAEE6), SPH_C64(0x92722C851482353B),
-	SPH_C64(0xA2BFE8A14CF10364), SPH_C64(0xA81A664BBC423001),
-	SPH_C64(0xC24B8B70D0F89791), SPH_C64(0xC76C51A30654BE30),
-	SPH_C64(0xD192E819D6EF5218), SPH_C64(0xD69906245565A910),
-	SPH_C64(0xF40E35855771202A), SPH_C64(0x106AA07032BBD1B8),
-	SPH_C64(0x19A4C116B8D2D0C8), SPH_C64(0x1E376C085141AB53),
-	SPH_C64(0x2748774CDF8EEB99), SPH_C64(0x34B0BCB5E19B48A8),
-	SPH_C64(0x391C0CB3C5C95A63), SPH_C64(0x4ED8AA4AE3418ACB),
-	SPH_C64(0x5B9CCA4F7763E373), SPH_C64(0x682E6FF3D6B2B8A3),
-	SPH_C64(0x748F82EE5DEFB2FC), SPH_C64(0x78A5636F43172F60),
-	SPH_C64(0x84C87814A1F0AB72), SPH_C64(0x8CC702081A6439EC),
-	SPH_C64(0x90BEFFFA23631E28), SPH_C64(0xA4506CEBDE82BDE9),
-	SPH_C64(0xBEF9A3F7B2C67915), SPH_C64(0xC67178F2E372532B),
-	SPH_C64(0xCA273ECEEA26619C), SPH_C64(0xD186B8C721C0C207),
-	SPH_C64(0xEADA7DD6CDE0EB1E), SPH_C64(0xF57D4F7FEE6ED178),
-	SPH_C64(0x06F067AA72176FBA), SPH_C64(0x0A637DC5A2C898A6),
-	SPH_C64(0x113F9804BEF90DAE), SPH_C64(0x1B710B35131C471B),
-	SPH_C64(0x28DB77F523047D84), SPH_C64(0x32CAAB7B40C72493),
-	SPH_C64(0x3C9EBE0A15C9BEBC), SPH_C64(0x431D67C49C100D4C),
-	SPH_C64(0x4CC5D4BECB3E42B6), SPH_C64(0x597F299CFC657E2A),
-	SPH_C64(0x5FCB6FAB3AD6FAEC), SPH_C64(0x6C44198C4A475817)
-};
-
-#define CH(X, Y, Z) \
-   _mm256_xor_si256( _mm256_and_si256( _mm256_xor_si256( Y, Z ), X ), Z ) 
-
-#define MAJ(X, Y, Z) \
-   _mm256_or_si256( _mm256_and_si256( X, Y ), \
-                    _mm256_and_si256( _mm256_or_si256( X, Y ), Z ) )
-
-#define BSG5_0(x) \
-   _mm256_xor_si256( _mm256_xor_si256( \
-        mm256_ror_64(x, 28), mm256_ror_64(x, 34) ), mm256_ror_64(x, 39) )
-
-#define BSG5_1(x) \
-   _mm256_xor_si256( _mm256_xor_si256( \
-        mm256_ror_64(x, 14), mm256_ror_64(x, 18) ), mm256_ror_64(x, 41) )
-
-#define SSG5_0(x) \
-   _mm256_xor_si256( _mm256_xor_si256( \
-        mm256_ror_64(x,  1), mm256_ror_64(x,  8) ), _mm256_srli_epi64(x, 7) ) 
-
-#define SSG5_1(x) \
-   _mm256_xor_si256( _mm256_xor_si256( \
-        mm256_ror_64(x, 19), mm256_ror_64(x, 61) ), _mm256_srli_epi64(x, 6) )
-
-#define SHA3_4WAY_STEP(A, B, C, D, E, F, G, H, i) \
-do { \
-  register __m256i T1, T2; \
-  T1 = _mm256_add_epi64( H, mm256_add4_64( BSG5_1(E), CH(E, F, G), \
-                                _mm256_set1_epi64x( K512[i] ), W[i] ) ); \
-  T2 = _mm256_add_epi64( BSG5_0(A), MAJ(A, B, C) ); \
-  D  = _mm256_add_epi64( D, T1 ); \
-  H  = _mm256_add_epi64( T1, T2 ); \
-} while (0)
-
-static void
-sha512_4way_round( __m256i *in, __m256i r[8] )
-{
-   int i;
-   register __m256i A, B, C, D, E, F, G, H;
-   __m256i W[80];
-
-   mm256_block_bswap_64( W  , in );
-   mm256_block_bswap_64( W+8, in+8 );
-
-   for ( i = 16; i < 80; i++ )
-      W[i] = mm256_add4_64( SSG5_1( W[ i- 2 ] ), W[ i- 7 ],
-                            SSG5_0( W[ i-15 ] ), W[ i-16 ] );
-
-   A = r[0];
-   B = r[1];
-   C = r[2];
-   D = r[3];
-   E = r[4];
-   F = r[5];
-   G = r[6];
-   H = r[7];
-
-   for ( i = 0; i < 80; i += 8 )
-   {
-      SHA3_4WAY_STEP( A, B, C, D, E, F, G, H, i + 0 );
-      SHA3_4WAY_STEP( H, A, B, C, D, E, F, G, i + 1 );
-      SHA3_4WAY_STEP( G, H, A, B, C, D, E, F, i + 2 );
-      SHA3_4WAY_STEP( F, G, H, A, B, C, D, E, i + 3 );
-      SHA3_4WAY_STEP( E, F, G, H, A, B, C, D, i + 4 );
-      SHA3_4WAY_STEP( D, E, F, G, H, A, B, C, i + 5 );
-      SHA3_4WAY_STEP( C, D, E, F, G, H, A, B, i + 6 );
-      SHA3_4WAY_STEP( B, C, D, E, F, G, H, A, i + 7 );
-   }
-
-   r[0] = _mm256_add_epi64( r[0], A );
-   r[1] = _mm256_add_epi64( r[1], B );
-   r[2] = _mm256_add_epi64( r[2], C );
-   r[3] = _mm256_add_epi64( r[3], D );
-   r[4] = _mm256_add_epi64( r[4], E );
-   r[5] = _mm256_add_epi64( r[5], F );
-   r[6] = _mm256_add_epi64( r[6], G );
-   r[7] = _mm256_add_epi64( r[7], H );
-}
-
-void sha512_4way_init( sha512_4way_context *sc )
-{
-   sc->count = 0;
-   sc->val[0] = _mm256_set1_epi64x( H512[0] );
-   sc->val[1] = _mm256_set1_epi64x( H512[1] );
-   sc->val[2] = _mm256_set1_epi64x( H512[2] );
-   sc->val[3] = _mm256_set1_epi64x( H512[3] );
-   sc->val[4] = _mm256_set1_epi64x( H512[4] );
-   sc->val[5] = _mm256_set1_epi64x( H512[5] );
-   sc->val[6] = _mm256_set1_epi64x( H512[6] );
-   sc->val[7] = _mm256_set1_epi64x( H512[7] );
-}
-
-void sha512_4way( sha512_4way_context *sc, const void *data, size_t len )
-{
-   __m256i *vdata = (__m256i*)data;
-   size_t ptr;
-   const int buf_size = 128;
-
-   ptr = (unsigned)sc->count & (buf_size - 1U);
-   while ( len > 0 )
-   {
-      size_t clen;
-      clen = buf_size - ptr;
-      if ( clen > len )
-         clen = len;
-      memcpy_256( sc->buf + (ptr>>3), vdata, clen>>3 );
-      vdata = vdata + (clen>>3);
-      ptr += clen;
-      len -= clen;
-      if ( ptr == buf_size )
-      {
-         sha512_4way_round( sc->buf, sc->val );
-         ptr = 0;
-      }
-      sc->count += clen;
-   }
-}
-
-void sha512_4way_close( sha512_4way_context *sc, void *dst )
-{
-    unsigned ptr;
-    const int buf_size = 128;
-    const int pad = buf_size - 16;
-
-    ptr = (unsigned)sc->count & (buf_size - 1U);
-    sc->buf[ ptr>>3 ] = m256_const1_64( 0x80 );
-    ptr += 8;
-    if ( ptr > pad )
-    {
-         memset_zero_256( sc->buf + (ptr>>3), (buf_size - ptr) >> 3 );
-         sha512_4way_round( sc->buf, sc->val );
-         memset_zero_256( sc->buf, pad >> 3 );
-    }
-    else
-         memset_zero_256( sc->buf + (ptr>>3), (pad - ptr) >> 3 );
-
-    sc->buf[ pad >> 3 ] =
-                 mm256_bswap_64( _mm256_set1_epi64x( sc->count >> 61 ) );
-    sc->buf[ ( pad+8 ) >> 3 ] = 
-                 mm256_bswap_64( _mm256_set1_epi64x( sc->count << 3 ) );
-    sha512_4way_round( sc->buf, sc->val );
-
-    mm256_block_bswap_64( dst, sc->val );
-}
-
 #endif  // __AVX2__
 #endif  // __SSE2__

algo/sha/sha256q-4way.c

@@ -3,7 +3,7 @@
 #include <stdint.h>
 #include <string.h>
 #include <stdio.h>
-#include "sha2-hash-4way.h"
+#include "sha-hash-4way.h"
 
 #if defined(SHA256T_8WAY)
 

algo/sha/sha256t-4way.c

@@ -3,7 +3,7 @@
 #include <stdint.h>
 #include <string.h>
 #include <stdio.h>
-#include "sha2-hash-4way.h"
+#include "sha-hash-4way.h"
 
 #if defined(SHA256T_11WAY)
 
@@ -158,7 +158,7 @@ void sha256t_8way_hash( void* output, const void* input )
    sha256_8way_close( &ctx, output );
 }
 
-int scanhash_sha256t_8way( struct work *work, uint32_t max_nonce,
+int scanhash_sha256t_8way( struct work *work, const uint32_t max_nonce,
                            uint64_t *hashes_done, struct thr_info *mythr )
 {
    uint32_t vdata[20*8]  __attribute__ ((aligned (64)));
@@ -166,12 +166,12 @@ int scanhash_sha256t_8way( struct work *work, uint32_t max_nonce,
    uint32_t lane_hash[8] __attribute__ ((aligned (32)));
    uint32_t *hash7 = &(hash[7<<3]);
    uint32_t *pdata = work->data;
-   uint32_t *ptarget = work->target;
+   const uint32_t *ptarget = work->target;
    const uint32_t Htarg = ptarget[7];
    const uint32_t first_nonce = pdata[19];
    uint32_t n = first_nonce;
    __m256i  *noncev = (__m256i*)vdata + 19;   // aligned
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id;
 
    const uint64_t htmax[] = {          0,
                                      0xF,
@@ -194,7 +194,7 @@ int scanhash_sha256t_8way( struct work *work, uint32_t max_nonce,
 
    for ( int m = 0; m < 6; m++ ) if ( Htarg <= htmax[m] )
    {
-      uint32_t mask = masks[m];
+      const uint32_t mask = masks[m];
       do
       {
         *noncev = mm256_bswap_32( _mm256_set_epi32(
@@ -244,7 +244,7 @@ void sha256t_4way_hash( void* output, const void* input )
    sha256_4way_close( &ctx, output );
 }
 
-int scanhash_sha256t_4way( struct work *work, uint32_t max_nonce,
+int scanhash_sha256t_4way( struct work *work, const uint32_t max_nonce,
 	                   uint64_t *hashes_done, struct thr_info *mythr )
 {
    uint32_t vdata[20*4] __attribute__ ((aligned (64)));
@@ -252,12 +252,12 @@ int scanhash_sha256t_4way( struct work *work, uint32_t max_nonce,
    uint32_t lane_hash[8] __attribute__ ((aligned (64)));
    uint32_t *hash7 = &(hash[7<<2]);
    uint32_t *pdata = work->data;
-   uint32_t *ptarget = work->target;
+   const uint32_t *ptarget = work->target;
    const uint32_t Htarg = ptarget[7];
    const uint32_t first_nonce = pdata[19];
    uint32_t n = first_nonce;
    __m128i  *noncev = (__m128i*)vdata + 19;   // aligned
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id;
 
    const uint64_t htmax[] = {          0,
                                      0xF,
@@ -278,7 +278,7 @@ int scanhash_sha256t_4way( struct work *work, uint32_t max_nonce,
 
    for ( int m = 0; m < 6; m++ ) if ( Htarg <= htmax[m] )
    {
-      uint32_t mask = masks[m];
+      const uint32_t mask = masks[m];
       do {
          *noncev = mm128_bswap_32( _mm_set_epi32( n+3,n+2,n+1,n ) );
          pdata[19] = n;

algo/sha/sha256t-gate.c

@@ -15,7 +15,6 @@ bool register_sha256t_algo( algo_gate_t* gate )
     gate->scanhash   = (void*)&scanhash_sha256t;
     gate->hash       = (void*)&sha256t_hash;
 #endif
-    gate->get_max64  = (void*)&get_max64_0x3ffff;
     return true;
 }
 
@@ -34,7 +33,6 @@ bool register_sha256q_algo( algo_gate_t* gate )
     gate->scanhash   = (void*)&scanhash_sha256q;
     gate->hash       = (void*)&sha256q_hash;
 #endif
-    gate->get_max64  = (void*)&get_max64_0x3ffff;
     return true;
 
 }

algo/sha/sha512-hash-4way.c

@@ -0,0 +1,312 @@
+/* $Id: sha2big.c 216 2010-06-08 09:46:57Z tp $ */
+/*
+ * SHA-384 / SHA-512 implementation.
+ *
+ * ==========================(LICENSE BEGIN)============================
+ *
+ * Copyright (c) 2007-2010  Projet RNRT SAPHIR
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+ * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * ===========================(LICENSE END)=============================
+ *
+ * @author   Thomas Pornin <thomas.pornin@cryptolog.com>
+ */
+
+#if defined(__AVX2__)
+
+#include <stddef.h>
+#include <string.h>
+#include "sha-hash-4way.h"
+
+// SHA-512 4 way 64 bit
+
+/*
+static const sph_u64 H512[8] = {
+        SPH_C64(0x6A09E667F3BCC908), SPH_C64(0xBB67AE8584CAA73B),
+        SPH_C64(0x3C6EF372FE94F82B), SPH_C64(0xA54FF53A5F1D36F1),
+        SPH_C64(0x510E527FADE682D1), SPH_C64(0x9B05688C2B3E6C1F),
+        SPH_C64(0x1F83D9ABFB41BD6B), SPH_C64(0x5BE0CD19137E2179)
+};
+*/
+
+static const sph_u64 K512[80] = {
+	SPH_C64(0x428A2F98D728AE22), SPH_C64(0x7137449123EF65CD),
+	SPH_C64(0xB5C0FBCFEC4D3B2F), SPH_C64(0xE9B5DBA58189DBBC),
+	SPH_C64(0x3956C25BF348B538), SPH_C64(0x59F111F1B605D019),
+	SPH_C64(0x923F82A4AF194F9B), SPH_C64(0xAB1C5ED5DA6D8118),
+	SPH_C64(0xD807AA98A3030242), SPH_C64(0x12835B0145706FBE),
+	SPH_C64(0x243185BE4EE4B28C), SPH_C64(0x550C7DC3D5FFB4E2),
+	SPH_C64(0x72BE5D74F27B896F), SPH_C64(0x80DEB1FE3B1696B1),
+	SPH_C64(0x9BDC06A725C71235), SPH_C64(0xC19BF174CF692694),
+	SPH_C64(0xE49B69C19EF14AD2), SPH_C64(0xEFBE4786384F25E3),
+	SPH_C64(0x0FC19DC68B8CD5B5), SPH_C64(0x240CA1CC77AC9C65),
+	SPH_C64(0x2DE92C6F592B0275), SPH_C64(0x4A7484AA6EA6E483),
+	SPH_C64(0x5CB0A9DCBD41FBD4), SPH_C64(0x76F988DA831153B5),
+	SPH_C64(0x983E5152EE66DFAB), SPH_C64(0xA831C66D2DB43210),
+	SPH_C64(0xB00327C898FB213F), SPH_C64(0xBF597FC7BEEF0EE4),
+	SPH_C64(0xC6E00BF33DA88FC2), SPH_C64(0xD5A79147930AA725),
+	SPH_C64(0x06CA6351E003826F), SPH_C64(0x142929670A0E6E70),
+	SPH_C64(0x27B70A8546D22FFC), SPH_C64(0x2E1B21385C26C926),
+	SPH_C64(0x4D2C6DFC5AC42AED), SPH_C64(0x53380D139D95B3DF),
+	SPH_C64(0x650A73548BAF63DE), SPH_C64(0x766A0ABB3C77B2A8),
+	SPH_C64(0x81C2C92E47EDAEE6), SPH_C64(0x92722C851482353B),
+	SPH_C64(0xA2BFE8A14CF10364), SPH_C64(0xA81A664BBC423001),
+	SPH_C64(0xC24B8B70D0F89791), SPH_C64(0xC76C51A30654BE30),
+	SPH_C64(0xD192E819D6EF5218), SPH_C64(0xD69906245565A910),
+	SPH_C64(0xF40E35855771202A), SPH_C64(0x106AA07032BBD1B8),
+	SPH_C64(0x19A4C116B8D2D0C8), SPH_C64(0x1E376C085141AB53),
+	SPH_C64(0x2748774CDF8EEB99), SPH_C64(0x34B0BCB5E19B48A8),
+	SPH_C64(0x391C0CB3C5C95A63), SPH_C64(0x4ED8AA4AE3418ACB),
+	SPH_C64(0x5B9CCA4F7763E373), SPH_C64(0x682E6FF3D6B2B8A3),
+	SPH_C64(0x748F82EE5DEFB2FC), SPH_C64(0x78A5636F43172F60),
+	SPH_C64(0x84C87814A1F0AB72), SPH_C64(0x8CC702081A6439EC),
+	SPH_C64(0x90BEFFFA23631E28), SPH_C64(0xA4506CEBDE82BDE9),
+	SPH_C64(0xBEF9A3F7B2C67915), SPH_C64(0xC67178F2E372532B),
+	SPH_C64(0xCA273ECEEA26619C), SPH_C64(0xD186B8C721C0C207),
+	SPH_C64(0xEADA7DD6CDE0EB1E), SPH_C64(0xF57D4F7FEE6ED178),
+	SPH_C64(0x06F067AA72176FBA), SPH_C64(0x0A637DC5A2C898A6),
+	SPH_C64(0x113F9804BEF90DAE), SPH_C64(0x1B710B35131C471B),
+	SPH_C64(0x28DB77F523047D84), SPH_C64(0x32CAAB7B40C72493),
+	SPH_C64(0x3C9EBE0A15C9BEBC), SPH_C64(0x431D67C49C100D4C),
+	SPH_C64(0x4CC5D4BECB3E42B6), SPH_C64(0x597F299CFC657E2A),
+	SPH_C64(0x5FCB6FAB3AD6FAEC), SPH_C64(0x6C44198C4A475817)
+};
+
+#define CH(X, Y, Z) \
+   _mm256_xor_si256( _mm256_and_si256( _mm256_xor_si256( Y, Z ), X ), Z ) 
+
+#define MAJ(X, Y, Z) \
+   _mm256_or_si256( _mm256_and_si256( X, Y ), \
+                    _mm256_and_si256( _mm256_or_si256( X, Y ), Z ) )
+
+#define BSG5_0(x) \
+   _mm256_xor_si256( _mm256_xor_si256( \
+        mm256_ror_64(x, 28), mm256_ror_64(x, 34) ), mm256_ror_64(x, 39) )
+
+#define BSG5_1(x) \
+   _mm256_xor_si256( _mm256_xor_si256( \
+        mm256_ror_64(x, 14), mm256_ror_64(x, 18) ), mm256_ror_64(x, 41) )
+
+#define SSG5_0(x) \
+   _mm256_xor_si256( _mm256_xor_si256( \
+        mm256_ror_64(x,  1), mm256_ror_64(x,  8) ), _mm256_srli_epi64(x, 7) ) 
+
+#define SSG5_1(x) \
+   _mm256_xor_si256( _mm256_xor_si256( \
+        mm256_ror_64(x, 19), mm256_ror_64(x, 61) ), _mm256_srli_epi64(x, 6) )
+
+// Interleave SSG0 & SSG1 for better throughput.
+// return ssg0(w0) + ssg1(w1)
+static inline __m256i ssg512_add( __m256i w0, __m256i w1 )
+{
+   __m256i w0a, w1a, w0b, w1b;
+   w0a = mm256_ror_64( w0, 1 );
+   w1a = mm256_ror_64( w1,19 );
+   w0b = mm256_ror_64( w0, 8 );
+   w1b = mm256_ror_64( w1,61 );
+   w0a = _mm256_xor_si256( w0a, w0b );
+   w1a = _mm256_xor_si256( w1a, w1b );
+   w0b = _mm256_srli_epi64( w0, 7 );
+   w1b = _mm256_srli_epi64( w1, 6 );
+   w0a = _mm256_xor_si256( w0a, w0b );
+   w1a = _mm256_xor_si256( w1a, w1b );
+   return _mm256_add_epi64( w0a, w1a );
+}
+
+
+#define SSG512x2_0( w0, w1, i ) do \
+{ \
+   __m256i X0a, X1a, X0b, X1b; \
+  X0a = mm256_ror_64( W[i-15], 1 ); \
+  X1a = mm256_ror_64( W[i-14], 1 ); \
+  X0b = mm256_ror_64( W[i-15], 8 ); \
+  X1b = mm256_ror_64( W[i-14], 8 ); \
+  X0a = _mm256_xor_si256( X0a, X0b ); \
+  X1a = _mm256_xor_si256( X1a, X1b ); \
+  X0b = _mm256_srli_epi64( W[i-15], 7 ); \
+  X1b = _mm256_srli_epi64( W[i-14], 7 ); \
+  w0  = _mm256_xor_si256( X0a, X0b ); \
+  w1  = _mm256_xor_si256( X1a, X1b ); \
+} while(0)
+
+#define SSG512x2_1( w0, w1, i ) do \
+{ \
+   __m256i X0a, X1a, X0b, X1b; \
+  X0a = mm256_ror_64( W[i-2],19 ); \
+  X1a = mm256_ror_64( W[i-1],19 ); \
+  X0b = mm256_ror_64( W[i-2],61 ); \
+  X1b = mm256_ror_64( W[i-1],61 ); \
+  X0a = _mm256_xor_si256( X0a, X0b ); \
+  X1a = _mm256_xor_si256( X1a, X1b ); \
+  X0b = _mm256_srli_epi64( W[i-2], 6 ); \
+  X1b = _mm256_srli_epi64( W[i-1], 6 ); \
+  w0  = _mm256_xor_si256( X0a, X0b ); \
+  w1  = _mm256_xor_si256( X1a, X1b ); \
+} while(0)
+
+#define SHA3_4WAY_STEP(A, B, C, D, E, F, G, H, i) \
+do { \
+  __m256i T1, T2; \
+  __m256i K = _mm256_set1_epi64x( K512[ i ] ); \
+  T1 = _mm256_add_epi64( H, mm256_add4_64( BSG5_1(E), CH(E, F, G), \
+                                           K, W[i] ) ); \
+  T2 = _mm256_add_epi64( BSG5_0(A), MAJ(A, B, C) ); \
+  D  = _mm256_add_epi64( D, T1 ); \
+  H  = _mm256_add_epi64( T1, T2 ); \
+} while (0)
+
+
+static void
+sha512_4way_round( sha512_4way_context *ctx,  __m256i *in, __m256i r[8] )
+{
+   int i;
+   register __m256i A, B, C, D, E, F, G, H;
+   __m256i W[80];
+
+   mm256_block_bswap_64( W  , in );
+   mm256_block_bswap_64( W+8, in+8 );
+
+   for ( i = 16; i < 80; i++ )
+      W[i] = _mm256_add_epi64( ssg512_add( W[i-15], W[i-2] ),
+                               _mm256_add_epi64( W[ i- 7 ], W[ i-16 ] ) );
+
+   if ( ctx->initialized )
+   {
+      A = r[0];
+      B = r[1];
+      C = r[2];
+      D = r[3];
+      E = r[4];
+      F = r[5];
+      G = r[6];
+      H = r[7];
+   }
+   else
+   {
+      A = m256_const1_64( 0x6A09E667F3BCC908 );
+      B = m256_const1_64( 0xBB67AE8584CAA73B );
+      C = m256_const1_64( 0x3C6EF372FE94F82B );
+      D = m256_const1_64( 0xA54FF53A5F1D36F1 );
+      E = m256_const1_64( 0x510E527FADE682D1 );
+      F = m256_const1_64( 0x9B05688C2B3E6C1F );
+      G = m256_const1_64( 0x1F83D9ABFB41BD6B );
+      H = m256_const1_64( 0x5BE0CD19137E2179 );
+   }
+
+   for ( i = 0; i < 80; i += 8 )
+   {
+      SHA3_4WAY_STEP( A, B, C, D, E, F, G, H, i + 0 );
+      SHA3_4WAY_STEP( H, A, B, C, D, E, F, G, i + 1 );
+      SHA3_4WAY_STEP( G, H, A, B, C, D, E, F, i + 2 );
+      SHA3_4WAY_STEP( F, G, H, A, B, C, D, E, i + 3 );
+      SHA3_4WAY_STEP( E, F, G, H, A, B, C, D, i + 4 );
+      SHA3_4WAY_STEP( D, E, F, G, H, A, B, C, i + 5 );
+      SHA3_4WAY_STEP( C, D, E, F, G, H, A, B, i + 6 );
+      SHA3_4WAY_STEP( B, C, D, E, F, G, H, A, i + 7 );
+   }
+
+   if ( ctx->initialized )
+   {
+      r[0] = _mm256_add_epi64( r[0], A );
+      r[1] = _mm256_add_epi64( r[1], B );
+      r[2] = _mm256_add_epi64( r[2], C );
+      r[3] = _mm256_add_epi64( r[3], D );
+      r[4] = _mm256_add_epi64( r[4], E );
+      r[5] = _mm256_add_epi64( r[5], F );
+      r[6] = _mm256_add_epi64( r[6], G );
+      r[7] = _mm256_add_epi64( r[7], H );
+   }
+   else
+   {
+      ctx->initialized = true;
+      r[0] = _mm256_add_epi64( A, m256_const1_64( 0x6A09E667F3BCC908 ) );
+      r[1] = _mm256_add_epi64( B, m256_const1_64( 0xBB67AE8584CAA73B ) );
+      r[2] = _mm256_add_epi64( C, m256_const1_64( 0x3C6EF372FE94F82B ) );
+      r[3] = _mm256_add_epi64( D, m256_const1_64( 0xA54FF53A5F1D36F1 ) );
+      r[4] = _mm256_add_epi64( E, m256_const1_64( 0x510E527FADE682D1 ) );
+      r[5] = _mm256_add_epi64( F, m256_const1_64( 0x9B05688C2B3E6C1F ) );
+      r[6] = _mm256_add_epi64( G, m256_const1_64( 0x1F83D9ABFB41BD6B ) );
+      r[7] = _mm256_add_epi64( H, m256_const1_64( 0x5BE0CD19137E2179 ) );
+   }
+}
+
+void sha512_4way_init( sha512_4way_context *sc )
+{
+   sc->initialized = false;
+   sc->count = 0;
+}
+
+void sha512_4way( sha512_4way_context *sc, const void *data, size_t len )
+{
+   __m256i *vdata = (__m256i*)data;
+   size_t ptr;
+   const int buf_size = 128;
+
+   ptr = (unsigned)sc->count & (buf_size - 1U);
+   while ( len > 0 )
+   {
+      size_t clen;
+      clen = buf_size - ptr;
+      if ( clen > len )
+         clen = len;
+      memcpy_256( sc->buf + (ptr>>3), vdata, clen>>3 );
+      vdata = vdata + (clen>>3);
+      ptr += clen;
+      len -= clen;
+      if ( ptr == buf_size )
+      {
+         sha512_4way_round( sc, sc->buf, sc->val );
+         ptr = 0;
+      }
+      sc->count += clen;
+   }
+}
+
+void sha512_4way_close( sha512_4way_context *sc, void *dst )
+{
+    unsigned ptr;
+    const int buf_size = 128;
+    const int pad = buf_size - 16;
+    const __m256i shuff_bswap64 = m256_const2_64( 0x08090a0b0c0d0e0f,
+                                                  0x0001020304050607 );
+
+    ptr = (unsigned)sc->count & (buf_size - 1U);
+    sc->buf[ ptr>>3 ] = m256_const1_64( 0x80 );
+    ptr += 8;
+    if ( ptr > pad )
+    {
+         memset_zero_256( sc->buf + (ptr>>3), (buf_size - ptr) >> 3 );
+         sha512_4way_round( sc, sc->buf, sc->val );
+         memset_zero_256( sc->buf, pad >> 3 );
+    }
+    else
+         memset_zero_256( sc->buf + (ptr>>3), (pad - ptr) >> 3 );
+
+    sc->buf[ pad >> 3 ] = _mm256_shuffle_epi8(
+                       _mm256_set1_epi64x( sc->count >> 61 ), shuff_bswap64 );
+    sc->buf[ ( pad+8 ) >> 3 ] = _mm256_shuffle_epi8( 
+                       _mm256_set1_epi64x( sc->count <<  3 ), shuff_bswap64 );
+    sha512_4way_round( sc, sc->buf, sc->val );
+
+    mm256_block_bswap_64( dst, sc->val );
+}
+
+#endif  // __AVX2__

algo/shabal/shabal-hash-4way.c

@@ -63,7 +63,6 @@ extern "C"{
  * that it can optimize them at will.
  */
 
-/* BEGIN -- automatically generated code. */
 
 #define DECL_STATE   \
 	__m128i A00, A01, A02, A03, A04, A05, A06, A07, \
@@ -76,8 +75,11 @@ extern "C"{
 	        M8, M9, MA, MB, MC, MD, ME, MF; \
 	sph_u32 Wlow, Whigh;
 
-#define READ_STATE(state)   do { \
-		A00 = (state)->A[0]; \
+#define READ_STATE(state) do \
+{ \
+   if ( (state)->state_loaded ) \
+   { \
+      A00 = (state)->A[0]; \
 		A01 = (state)->A[1]; \
 		A02 = (state)->A[2]; \
 		A03 = (state)->A[3]; \
@@ -121,9 +123,58 @@ extern "C"{
 		CD = (state)->C[13]; \
 		CE = (state)->C[14]; \
 		CF = (state)->C[15]; \
-		Wlow = (state)->Wlow; \
-		Whigh = (state)->Whigh; \
-	} while (0)
+   } \
+   else \
+   { \
+       (state)->state_loaded = true; \
+       A00 = m128_const1_64( 0x20728DFD20728DFD ); \
+       A01 = m128_const1_64( 0x46C0BD5346C0BD53 ); \
+       A02 = m128_const1_64( 0xE782B699E782B699 ); \
+       A03 = m128_const1_64( 0x5530463255304632 ); \
+       A04 = m128_const1_64( 0x71B4EF9071B4EF90 ); \
+       A05 = m128_const1_64( 0x0EA9E82C0EA9E82C ); \
+       A06 = m128_const1_64( 0xDBB930F1DBB930F1 ); \
+       A07 = m128_const1_64( 0xFAD06B8BFAD06B8B ); \
+       A08 = m128_const1_64( 0xBE0CAE40BE0CAE40 ); \
+       A09 = m128_const1_64( 0x8BD144108BD14410 ); \
+       A0A = m128_const1_64( 0x76D2ADAC76D2ADAC ); \
+       A0B = m128_const1_64( 0x28ACAB7F28ACAB7F ); \
+       B0 = m128_const1_64( 0xC1099CB7C1099CB7 ); \
+       B1 = m128_const1_64( 0x07B385F307B385F3 ); \
+       B2 = m128_const1_64( 0xE7442C26E7442C26 ); \
+       B3 = m128_const1_64( 0xCC8AD640CC8AD640 ); \
+       B4 = m128_const1_64( 0xEB6F56C7EB6F56C7 ); \
+       B5 = m128_const1_64( 0x1EA81AA91EA81AA9 ); \
+       B6 = m128_const1_64( 0x73B9D31473B9D314 ); \
+       B7 = m128_const1_64( 0x1DE85D081DE85D08 ); \
+       B8 = m128_const1_64( 0x48910A5A48910A5A ); \
+       B9 = m128_const1_64( 0x893B22DB893B22DB ); \
+       BA = m128_const1_64( 0xC5A0DF44C5A0DF44 ); \
+       BB = m128_const1_64( 0xBBC4324EBBC4324E ); \
+       BC = m128_const1_64( 0x72D2F24072D2F240 ); \
+       BD = m128_const1_64( 0x75941D9975941D99 ); \
+       BE = m128_const1_64( 0x6D8BDE826D8BDE82 ); \
+       BF = m128_const1_64( 0xA1A7502BA1A7502B ); \
+       C0 = m128_const1_64( 0xD9BF68D1D9BF68D1 ); \
+       C1 = m128_const1_64( 0x58BAD75058BAD750 ); \
+       C2 = m128_const1_64( 0x56028CB256028CB2 ); \
+       C3 = m128_const1_64( 0x8134F3598134F359 ); \
+       C4 = m128_const1_64( 0xB5D469D8B5D469D8 ); \
+       C5 = m128_const1_64( 0x941A8CC2941A8CC2 ); \
+       C6 = m128_const1_64( 0x418B2A6E418B2A6E ); \
+       C7 = m128_const1_64( 0x0405278004052780 ); \
+       C8 = m128_const1_64( 0x7F07D7877F07D787 ); \
+       C9 = m128_const1_64( 0x5194358F5194358F ); \
+       CA = m128_const1_64( 0x3C60D6653C60D665 ); \
+       CB = m128_const1_64( 0xBE97D79ABE97D79A ); \
+       CC = m128_const1_64( 0x950C3434950C3434 ); \
+       CD = m128_const1_64( 0xAED9A06DAED9A06D ); \
+       CE = m128_const1_64( 0x2537DC8D2537DC8D ); \
+       CF = m128_const1_64( 0x7CDB59697CDB5969 ); \
+   } \
+   Wlow = (state)->Wlow; \
+   Whigh = (state)->Whigh; \
+} while (0)
 
 #define WRITE_STATE(state)   do { \
 		(state)->A[0] = A00; \
@@ -397,6 +448,7 @@ do { \
 			Whigh = T32(Whigh + 1); \
 	} while (0)
 
+/*
 static const sph_u32 A_init_256[] = {
 	C32(0x52F84552), C32(0xE54B7999), C32(0x2D8EE3EC), C32(0xB9645191),
 	C32(0xE0078B86), C32(0xBB7C44C9), C32(0xD2B5C1CA), C32(0xB0D2EB8C),
@@ -436,33 +488,115 @@ static const sph_u32 C_init_512[] = {
 	C32(0x7F07D787), C32(0x5194358F), C32(0x3C60D665), C32(0xBE97D79A),
 	C32(0x950C3434), C32(0xAED9A06D), C32(0x2537DC8D), C32(0x7CDB5969)
 };
+*/
 
 static void
 shabal_4way_init( void *cc, unsigned size )
 {
    shabal_4way_context *sc = (shabal_4way_context*)cc;
-   int i;
 
    if ( size == 512 )
-   {
-      for ( i = 0; i < 12; i++ )
-         sc->A[i] = _mm_set1_epi32( A_init_512[i] );
-      for ( i = 0; i < 16; i++ )
-      {
-         sc->B[i] = _mm_set1_epi32( B_init_512[i] );
-         sc->C[i] = _mm_set1_epi32( C_init_512[i] );
-      }
+   { // copy immediate constants directly to working registers later.
+       sc->state_loaded = false;
+/*
+       sc->A[ 0] = m128_const1_64( 0x20728DFD20728DFD );
+       sc->A[ 1] = m128_const1_64( 0x46C0BD5346C0BD53 );
+       sc->A[ 2] = m128_const1_64( 0xE782B699E782B699 );
+       sc->A[ 3] = m128_const1_64( 0x5530463255304632 );
+       sc->A[ 4] = m128_const1_64( 0x71B4EF9071B4EF90 );
+       sc->A[ 5] = m128_const1_64( 0x0EA9E82C0EA9E82C );
+       sc->A[ 6] = m128_const1_64( 0xDBB930F1DBB930F1 );
+       sc->A[ 7] = m128_const1_64( 0xFAD06B8BFAD06B8B );
+       sc->A[ 8] = m128_const1_64( 0xBE0CAE40BE0CAE40 );
+       sc->A[ 9] = m128_const1_64( 0x8BD144108BD14410 );
+       sc->A[10] = m128_const1_64( 0x76D2ADAC76D2ADAC );
+       sc->A[11] = m128_const1_64( 0x28ACAB7F28ACAB7F );
+
+       sc->B[ 0] = m128_const1_64( 0xC1099CB7C1099CB7 );
+       sc->B[ 1] = m128_const1_64( 0x07B385F307B385F3 );
+       sc->B[ 2] = m128_const1_64( 0xE7442C26E7442C26 );
+       sc->B[ 3] = m128_const1_64( 0xCC8AD640CC8AD640 );
+       sc->B[ 4] = m128_const1_64( 0xEB6F56C7EB6F56C7 );
+       sc->B[ 5] = m128_const1_64( 0x1EA81AA91EA81AA9 );
+       sc->B[ 6] = m128_const1_64( 0x73B9D31473B9D314 );
+       sc->B[ 7] = m128_const1_64( 0x1DE85D081DE85D08 );
+       sc->B[ 8] = m128_const1_64( 0x48910A5A48910A5A );
+       sc->B[ 9] = m128_const1_64( 0x893B22DB893B22DB );
+       sc->B[10] = m128_const1_64( 0xC5A0DF44C5A0DF44 );
+       sc->B[11] = m128_const1_64( 0xBBC4324EBBC4324E );
+       sc->B[12] = m128_const1_64( 0x72D2F24072D2F240 );
+       sc->B[13] = m128_const1_64( 0x75941D9975941D99 );
+       sc->B[14] = m128_const1_64( 0x6D8BDE826D8BDE82 );
+       sc->B[15] = m128_const1_64( 0xA1A7502BA1A7502B );
+
+       sc->C[ 0] = m128_const1_64( 0xD9BF68D1D9BF68D1 );
+       sc->C[ 1] = m128_const1_64( 0x58BAD75058BAD750 );
+       sc->C[ 2] = m128_const1_64( 0x56028CB256028CB2 );
+       sc->C[ 3] = m128_const1_64( 0x8134F3598134F359 );
+       sc->C[ 4] = m128_const1_64( 0xB5D469D8B5D469D8 );
+       sc->C[ 5] = m128_const1_64( 0x941A8CC2941A8CC2 );
+       sc->C[ 6] = m128_const1_64( 0x418B2A6E418B2A6E );
+       sc->C[ 7] = m128_const1_64( 0x0405278004052780 );
+       sc->C[ 8] = m128_const1_64( 0x7F07D7877F07D787 );
+       sc->C[ 9] = m128_const1_64( 0x5194358F5194358F );
+       sc->C[10] = m128_const1_64( 0x3C60D6653C60D665 );
+       sc->C[11] = m128_const1_64( 0xBE97D79ABE97D79A );
+       sc->C[12] = m128_const1_64( 0x950C3434950C3434 );
+       sc->C[13] = m128_const1_64( 0xAED9A06DAED9A06D );
+       sc->C[14] = m128_const1_64( 0x2537DC8D2537DC8D );
+       sc->C[15] = m128_const1_64( 0x7CDB59697CDB5969 );
+*/
    }
    else
-   {
-      for ( i = 0; i < 12; i++ )
-         sc->A[i] = _mm_set1_epi32( A_init_256[i] );
-      for ( i = 0; i < 16; i++ )
-      {
-         sc->B[i] = _mm_set1_epi32( B_init_256[i] );
-         sc->C[i] = _mm_set1_epi32( C_init_256[i] );
-      }
-    }
+   {  // No users
+       sc->state_loaded = true;
+       sc->A[ 0] = m128_const1_64( 0x52F8455252F84552 );
+       sc->A[ 1] = m128_const1_64( 0xE54B7999E54B7999 );
+       sc->A[ 2] = m128_const1_64( 0x2D8EE3EC2D8EE3EC );
+       sc->A[ 3] = m128_const1_64( 0xB9645191B9645191 );
+       sc->A[ 4] = m128_const1_64( 0xE0078B86E0078B86 );
+       sc->A[ 5] = m128_const1_64( 0xBB7C44C9BB7C44C9 );
+       sc->A[ 6] = m128_const1_64( 0xD2B5C1CAD2B5C1CA );
+       sc->A[ 7] = m128_const1_64( 0xB0D2EB8CB0D2EB8C );
+       sc->A[ 8] = m128_const1_64( 0x14CE5A4514CE5A45 );
+       sc->A[ 9] = m128_const1_64( 0x22AF50DC22AF50DC );
+       sc->A[10] = m128_const1_64( 0xEFFDBC6BEFFDBC6B );
+       sc->A[11] = m128_const1_64( 0xEB21B74AEB21B74A );
+
+       sc->B[ 0] = m128_const1_64( 0xB555C6EEB555C6EE );
+       sc->B[ 1] = m128_const1_64( 0x3E7105963E710596 );
+       sc->B[ 2] = m128_const1_64( 0xA72A652FA72A652F );
+       sc->B[ 3] = m128_const1_64( 0x9301515F9301515F );
+       sc->B[ 4] = m128_const1_64( 0xDA28C1FADA28C1FA );
+       sc->B[ 5] = m128_const1_64( 0x696FD868696FD868 );
+       sc->B[ 6] = m128_const1_64( 0x9CB6BF729CB6BF72 );
+       sc->B[ 7] = m128_const1_64( 0x0AFE40020AFE4002 );
+       sc->B[ 8] = m128_const1_64( 0xA6E03615A6E03615 );
+       sc->B[ 9] = m128_const1_64( 0x5138C1D45138C1D4 );
+       sc->B[10] = m128_const1_64( 0xBE216306BE216306 );
+       sc->B[11] = m128_const1_64( 0xB38B8890B38B8890 );
+       sc->B[12] = m128_const1_64( 0x3EA8B96B3EA8B96B );
+       sc->B[13] = m128_const1_64( 0x3299ACE43299ACE4 );
+       sc->B[14] = m128_const1_64( 0x30924DD430924DD4 );
+       sc->B[15] = m128_const1_64( 0x55CB34A555CB34A5 );
+
+       sc->C[ 0] = m128_const1_64( 0xB405F031B405F031 );
+       sc->C[ 1] = m128_const1_64( 0xC4233EBAC4233EBA );
+       sc->C[ 2] = m128_const1_64( 0xB3733979B3733979 );
+       sc->C[ 3] = m128_const1_64( 0xC0DD9D55C0DD9D55 );
+       sc->C[ 4] = m128_const1_64( 0xC51C28AEC51C28AE );
+       sc->C[ 5] = m128_const1_64( 0xA327B8E1A327B8E1 );
+       sc->C[ 6] = m128_const1_64( 0x56C5616756C56167 );
+       sc->C[ 7] = m128_const1_64( 0xED614433ED614433 );
+       sc->C[ 8] = m128_const1_64( 0x88B59D6088B59D60 );
+       sc->C[ 9] = m128_const1_64( 0x60E2CEBA60E2CEBA );
+       sc->C[10] = m128_const1_64( 0x758B4B8B758B4B8B );
+       sc->C[11] = m128_const1_64( 0x83E82A7F83E82A7F );
+       sc->C[12] = m128_const1_64( 0xBC968828BC968828 );
+       sc->C[13] = m128_const1_64( 0xE6E00BF7E6E00BF7 );
+       sc->C[14] = m128_const1_64( 0xBA839E55BA839E55 );
+       sc->C[15] = m128_const1_64( 0x9B491C609B491C60 );
+   }
     sc->Wlow = 1;
     sc->Whigh = 0;
     sc->ptr = 0;
@@ -488,6 +622,8 @@ shabal_4way_core( void *cc, const unsigned char *data, size_t len )
       sc->ptr = ptr;
       return;
    }
+   
+
    READ_STATE(sc);
 
    while ( len > 0 )

algo/shabal/shabal-hash-4way.h

@@ -54,7 +54,8 @@ typedef struct {
 	__m128i buf[16] __attribute__ ((aligned (64)));
 	__m128i A[12], B[16], C[16];
 	sph_u32 Whigh, Wlow;
-        size_t ptr;
+   size_t ptr;
+   bool state_loaded;
 } shabal_4way_context;
 
 typedef shabal_4way_context shabal256_4way_context;

algo/shavite/shavite-hash-2way.c

@@ -5,6 +5,7 @@
 
 #if defined(__AVX2__)
 
+
 static const uint32_t IV512[] =
 {
         0x72FCCDD8, 0x79CA4727, 0x128A077B, 0x40D55AEC,
@@ -13,6 +14,7 @@ static const uint32_t IV512[] =
         0xE275EADE, 0x502D9FCD, 0xB9357178, 0x022A4B9A
 };
 
+
 #define mm256_ror2x256hi_1x32( a, b ) \
    _mm256_blend_epi32( mm256_ror1x32_128( a ), \
                        mm256_ror1x32_128( b ), 0x88 )
@@ -232,18 +234,14 @@ c512_2way( shavite512_2way_context *ctx, const void *msg )
 
 void shavite512_2way_init( shavite512_2way_context *ctx )
 {
-   casti_m256i( ctx->h, 0 ) =
-            _mm256_set_epi32( IV512[ 3], IV512[ 2], IV512[ 1], IV512[ 0],
-                              IV512[ 3], IV512[ 2], IV512[ 1], IV512[ 0] );  
-   casti_m256i( ctx->h, 1 ) =
-            _mm256_set_epi32( IV512[ 7], IV512[ 6], IV512[ 5], IV512[ 4],
-                              IV512[ 7], IV512[ 6], IV512[ 5], IV512[ 4] );
-   casti_m256i( ctx->h, 2 ) =
-            _mm256_set_epi32( IV512[11], IV512[10], IV512[ 9], IV512[ 8],
-                              IV512[11], IV512[10], IV512[ 9], IV512[ 8] );
-   casti_m256i( ctx->h, 3 ) =
-            _mm256_set_epi32( IV512[15], IV512[14], IV512[13], IV512[12],
-                              IV512[15], IV512[14], IV512[13], IV512[12] );
+    __m256i *h = (__m256i*)ctx->h;
+    __m128i *iv = (__m128i*)IV512;
+   
+   h[0] = m256_const1_128( iv[0] );
+   h[1] = m256_const1_128( iv[1] );
+   h[2] = m256_const1_128( iv[2] );
+   h[3] = m256_const1_128( iv[3] );
+
    ctx->ptr    = 0;
    ctx->count0 = 0;
    ctx->count1 = 0;
@@ -251,6 +249,7 @@ void shavite512_2way_init( shavite512_2way_context *ctx )
    ctx->count3 = 0;
 }
 
+// not tested, use update_close
 void shavite512_2way_update( shavite512_2way_context *ctx, const void *data,
                              size_t len )
 {
@@ -287,6 +286,7 @@ void shavite512_2way_update( shavite512_2way_context *ctx, const void *data,
    ctx->ptr = ptr;
 }
 
+// not tested
 void shavite512_2way_close( shavite512_2way_context *ctx, void *dst )
 {
     unsigned char *buf;
@@ -300,7 +300,7 @@ void shavite512_2way_close( shavite512_2way_context *ctx, void *dst )
     uint32_t vp = ctx->ptr>>5;
 
     // Terminating byte then zero pad
-    casti_m256i( buf, vp++ ) = _mm256_set_epi32( 0,0,0,0x80, 0,0,0,0x80 );
+    casti_m256i( buf, vp++ ) = m256_const2_64( 0, 0x0000000000000080 );
 
     // Zero pad full vectors up to count
     for ( ; vp < 6; vp++ )      
@@ -314,14 +314,12 @@ void shavite512_2way_close( shavite512_2way_context *ctx, void *dst )
     count.u32[2] = ctx->count2;
     count.u32[3] = ctx->count3;
 
-    casti_m256i( buf, 6 ) = _mm256_set_epi16( count.u16[0], 0,0,0,0,0,0,0,
-		                              count.u16[0], 0,0,0,0,0,0,0 );
-    casti_m256i( buf, 7 ) = _mm256_set_epi16(
-		    0x0200      , count.u16[7], count.u16[6], count.u16[5],
-		    count.u16[4], count.u16[3], count.u16[2], count.u16[1],
-                    0x0200      , count.u16[7], count.u16[6], count.u16[5],
-                    count.u16[4], count.u16[3], count.u16[2], count.u16[1] );
-
+    casti_m256i( buf, 6 ) = m256_const1_128(
+                  _mm_insert_epi16( m128_zero, count.u16[0], 7 ) ); 
+    casti_m256i( buf, 7 ) = m256_const1_128( _mm_set_epi16(
+                  0x0200,       count.u16[7], count.u16[6], count.u16[5],
+                  count.u16[4], count.u16[3], count.u16[2], count.u16[1] ) );
+                
     c512_2way( ctx, buf);
 
     casti_m256i( dst, 0 ) = casti_m256i( ctx->h, 0 );
@@ -382,23 +380,21 @@ void shavite512_2way_update_close( shavite512_2way_context *ctx, void *dst,
 
    if ( vp == 0 )    // empty buf, xevan.
    { 
-      casti_m256i( buf, 0 ) = _mm256_set_epi32( 0,0,0,0x80, 0,0,0,0x80 );
+      casti_m256i( buf, 0 ) = m256_const2_64( 0, 0x0000000000000080 );
       memset_zero_256( (__m256i*)buf + 1, 5 );
       ctx->count0 = ctx->count1 = ctx->count2 = ctx->count3 = 0;
    }
    else     // half full buf, everyone else.
    {
-      casti_m256i( buf, vp++ ) = _mm256_set_epi32( 0,0,0,0x80, 0,0,0,0x80 );
+    casti_m256i( buf, vp++ ) = m256_const2_64( 0, 0x0000000000000080 );
       memset_zero_256( (__m256i*)buf + vp, 6 - vp );
    }
 
-   casti_m256i( buf, 6 ) = _mm256_set_epi16( count.u16[0], 0,0,0,0,0,0,0,
-                                             count.u16[0], 0,0,0,0,0,0,0 );
-   casti_m256i( buf, 7 ) = _mm256_set_epi16(
-                   0x0200      , count.u16[7], count.u16[6], count.u16[5],
-                   count.u16[4], count.u16[3], count.u16[2], count.u16[1],
-                   0x0200      , count.u16[7], count.u16[6], count.u16[5],
-                   count.u16[4], count.u16[3], count.u16[2], count.u16[1] );
+    casti_m256i( buf, 6 ) = m256_const1_128(
+                  _mm_insert_epi16( m128_zero, count.u16[0], 7 ) ); 
+    casti_m256i( buf, 7 ) = m256_const1_128( _mm_set_epi16(
+                  0x0200,       count.u16[7], count.u16[6], count.u16[5],
+                  count.u16[4], count.u16[3], count.u16[2], count.u16[1] ) );
 
    c512_2way( ctx, buf);
 

algo/simd/simd-hash-2way.c

@@ -110,14 +110,26 @@ static const m256_v16 FFT256_Twiddle[] =
 
 // imported from vector.c
 
+#define REDUCE(x) \
+  _mm256_sub_epi16( _mm256_and_si256( x, m256_const1_64( \
+                         0x00ff00ff00ff00ff ) ), _mm256_srai_epi16( x, 8 ) )
+/*
 #define REDUCE(x) \
   _mm256_sub_epi16( _mm256_and_si256( x, _mm256_set1_epi16( 255 ) ), \
                                          _mm256_srai_epi16( x, 8 ) )
+*/
 
+#define EXTRA_REDUCE_S(x)\
+  _mm256_sub_epi16( x, _mm256_and_si256( \
+             m256_const1_64( 0x0101010101010101 ), \
+             _mm256_cmpgt_epi16( x, m256_const1_64( 0x0080008000800080 ) ) ) )
+
+/*
 #define EXTRA_REDUCE_S(x)\
   _mm256_sub_epi16( x, \
          _mm256_and_si256( _mm256_set1_epi16( 257 ), \
                            _mm256_cmpgt_epi16( x, _mm256_set1_epi16( 128 ) ) ) )
+*/
 
 #define REDUCE_FULL_S( x )  EXTRA_REDUCE_S( REDUCE (x ) )
 

algo/skein/skein-4way.c

@@ -5,7 +5,7 @@
 #if defined(__SHA__)
   #include <openssl/sha.h>
 #else
-  #include "algo/sha/sha2-hash-4way.h"
+  #include "algo/sha/sha-hash-4way.h"
 #endif
 
 #if defined (SKEIN_4WAY)

algo/skein/skein-gate.c

@@ -2,8 +2,6 @@
 #include "sph_skein.h"
 #include "skein-hash-4way.h"
 
-int64_t skein_get_max64() { return 0x7ffffLL; }
-
 bool register_skein_algo( algo_gate_t* gate )
 {
     gate->optimizations = AVX2_OPT | SHA_OPT;
@@ -14,7 +12,6 @@ bool register_skein_algo( algo_gate_t* gate )
     gate->scanhash  = (void*)&scanhash_skein;
     gate->hash      = (void*)&skeinhash;
 #endif
-    gate->get_max64 = (void*)&skein_get_max64;
     return true;
 };
 

algo/skein/skein-hash-4way.c

@@ -415,18 +415,46 @@ do { \
 		sc->bcount = bcount; \
 	} while (0)
 
+/*
+static const sph_u64 IV256[] = {
+   SPH_C64(0xCCD044A12FDB3E13), SPH_C64(0xE83590301A79A9EB),
+   SPH_C64(0x55AEA0614F816E6F), SPH_C64(0x2A2767A4AE9B94DB),
+   SPH_C64(0xEC06025E74DD7683), SPH_C64(0xE7A436CDC4746251),
+   SPH_C64(0xC36FBAF9393AD185), SPH_C64(0x3EEDBA1833EDFC13)
+};
 
-static void
-skein_big_init_4way( skein512_4way_context *sc, const sph_u64 *iv )
+static const sph_u64 IV512[] = {
+   SPH_C64(0x4903ADFF749C51CE), SPH_C64(0x0D95DE399746DF03),
+   SPH_C64(0x8FD1934127C79BCE), SPH_C64(0x9A255629FF352CB1),
+   SPH_C64(0x5DB62599DF6CA7B0), SPH_C64(0xEABE394CA9D5C3F4),
+   SPH_C64(0x991112C71A75B523), SPH_C64(0xAE18A40B660FCC33)
+};
+*/
+
+void skein256_4way_init( skein256_4way_context *sc )
 {
-        sc->h0 = _mm256_set_epi64x( iv[0], iv[0],iv[0],iv[0] );
-        sc->h1 = _mm256_set_epi64x( iv[1], iv[1],iv[1],iv[1] );
-        sc->h2 = _mm256_set_epi64x( iv[2], iv[2],iv[2],iv[2] );
-        sc->h3 = _mm256_set_epi64x( iv[3], iv[3],iv[3],iv[3] );
-        sc->h4 = _mm256_set_epi64x( iv[4], iv[4],iv[4],iv[4] );
-        sc->h5 = _mm256_set_epi64x( iv[5], iv[5],iv[5],iv[5] );
-        sc->h6 = _mm256_set_epi64x( iv[6], iv[6],iv[6],iv[6] );
-        sc->h7 = _mm256_set_epi64x( iv[7], iv[7],iv[7],iv[7] );
+        sc->h0 = m256_const1_64( 0xCCD044A12FDB3E13 );
+        sc->h1 = m256_const1_64( 0xE83590301A79A9EB );
+        sc->h2 = m256_const1_64( 0x55AEA0614F816E6F );
+        sc->h3 = m256_const1_64( 0x2A2767A4AE9B94DB );
+        sc->h4 = m256_const1_64( 0xEC06025E74DD7683 );
+        sc->h5 = m256_const1_64( 0xE7A436CDC4746251 );
+        sc->h6 = m256_const1_64( 0xC36FBAF9393AD185 );
+        sc->h7 = m256_const1_64( 0x3EEDBA1833EDFC13 );
+        sc->bcount = 0;
+        sc->ptr = 0;
+}
+
+void skein512_4way_init( skein512_4way_context *sc )
+{
+        sc->h0 = m256_const1_64( 0x4903ADFF749C51CE );
+        sc->h1 = m256_const1_64( 0x0D95DE399746DF03 );
+        sc->h2 = m256_const1_64( 0x8FD1934127C79BCE );
+        sc->h3 = m256_const1_64( 0x9A255629FF352CB1 );
+        sc->h4 = m256_const1_64( 0x5DB62599DF6CA7B0 );
+        sc->h5 = m256_const1_64( 0xEABE394CA9D5C3F4 );
+        sc->h6 = m256_const1_64( 0x991112C71A75B523 );
+        sc->h7 = m256_const1_64( 0xAE18A40B660FCC33 );
         sc->bcount = 0;
         sc->ptr = 0;
 }
@@ -524,6 +552,7 @@ skein_big_close_4way( skein512_4way_context *sc, unsigned ub, unsigned n,
         memcpy_256( dst, buf, out_len >> 3 );
 }
 
+/*
 static const sph_u64 IV256[] = {
 	SPH_C64(0xCCD044A12FDB3E13), SPH_C64(0xE83590301A79A9EB),
 	SPH_C64(0x55AEA0614F816E6F), SPH_C64(0x2A2767A4AE9B94DB),
@@ -537,13 +566,14 @@ static const sph_u64 IV512[] = {
 	SPH_C64(0x5DB62599DF6CA7B0), SPH_C64(0xEABE394CA9D5C3F4),
 	SPH_C64(0x991112C71A75B523), SPH_C64(0xAE18A40B660FCC33)
 };
-
-
+*/
+/*
 void
 skein256_4way_init(void *cc)
 {
 	skein_big_init_4way(cc, IV256);
 }
+*/
 
 void
 skein256_4way(void *cc, const void *data, size_t len)
@@ -557,11 +587,13 @@ skein256_4way_close(void *cc, void *dst)
         skein_big_close_4way(cc, 0, 0, dst, 32);
 }
 
+/*
 void
 skein512_4way_init(void *cc)
 {
 	skein_big_init_4way(cc, IV512);
 }
+*/
 
 void
 skein512_4way(void *cc, const void *data, size_t len)

algo/skein/skein-hash-4way.h

@@ -55,25 +55,26 @@ extern "C"{
 #define SPH_SIZE_skein256   256
 #define SPH_SIZE_skein512   512
 
-typedef struct {
-        __m256i buf[8] __attribute__ ((aligned (32)));
-        __m256i h0, h1, h2, h3, h4, h5, h6, h7;
-        size_t ptr;
+typedef struct
+{
+   __m256i buf[8] __attribute__ ((aligned (64)));
+   __m256i h0, h1, h2, h3, h4, h5, h6, h7;
+   size_t ptr;
 	sph_u64 bcount;
 } sph_skein_4way_big_context;
 
 typedef sph_skein_4way_big_context skein512_4way_context;
 typedef sph_skein_4way_big_context skein256_4way_context;
 
-void skein512_4way_init(void *cc);
-void skein512_4way(void *cc, const void *data, size_t len);
-void skein512_4way_close(void *cc, void *dst);
+void skein512_4way_init( skein512_4way_context *sc );
+void skein512_4way( void *cc, const void *data, size_t len );
+void skein512_4way_close( void *cc, void *dst );
 //void sph_skein512_addbits_and_close(
 //        void *cc, unsigned ub, unsigned n, void *dst);
 
-void skein256_4way_init(void *cc);
-void skein256_4way(void *cc, const void *data, size_t len);
-void skein256_4way_close(void *cc, void *dst);
+void skein256_4way_init( skein256_4way_context *sc );
+void skein256_4way( void *cc, const void *data, size_t len );
+void skein256_4way_close( void *cc, void *dst );
 //void sph_skein256_addbits_and_close(
 //	void *cc, unsigned ub, unsigned n, void *dst);
 

algo/skein/skein2-gate.c

@@ -2,11 +2,6 @@
 #include <stdint.h>
 #include "sph_skein.h"
 
-int64_t skein2_get_max64 ()
-{
-  return 0x7ffffLL;
-}
-
 bool register_skein2_algo( algo_gate_t* gate )
 {
   gate->optimizations = AVX2_OPT;
@@ -17,7 +12,6 @@ bool register_skein2_algo( algo_gate_t* gate )
   gate->scanhash  = (void*)&scanhash_skein2;
   gate->hash      = (void*)&skein2hash;
 #endif
-  gate->get_max64 = (void*)&skein2_get_max64;
   return true;
 };
 

algo/sm3/sm3-hash-4way.c

@@ -181,7 +181,7 @@ void sm3_4way_compress( __m128i *digest, __m128i *block )
    for( j =0; j < 16; j++ )
    {
       SS1 = mm128_rol_32( _mm_add_epi32( _mm_add_epi32( mm128_rol_32(A,12), E ),
-                                      mm128_rol_32( T, j ) ), 7 );
+                                      mm128_rol_var_32( T, j ) ), 7 );
       SS2 = _mm_xor_si128( SS1, mm128_rol_32( A, 12 ) );
       TT1 = _mm_add_epi32( _mm_add_epi32( _mm_add_epi32( FF0( A, B, C ), D ),
                                           SS2 ), W1[j] );
@@ -201,7 +201,7 @@ void sm3_4way_compress( __m128i *digest, __m128i *block )
    for( j =16; j < 64; j++ )
    {
       SS1 = mm128_rol_32( _mm_add_epi32( _mm_add_epi32( mm128_rol_32(A,12), E ),
-                                      mm128_rol_32( T, j&31 ) ), 7 );
+                                      mm128_rol_var_32( T, j&31 ) ), 7 );
       SS2 = _mm_xor_si128( SS1, mm128_rol_32( A, 12 ) );
       TT1 = _mm_add_epi32( _mm_add_epi32( _mm_add_epi32( FF1( A, B, C ), D ), 
                                           SS2 ), W1[j] );

algo/swifftx/Swifftx_sha3.cpp

@@ -0,0 +1,369 @@
+#include "Swifftx_sha3.h"
+extern "C" {
+#include "SWIFFTX.h"
+}
+#include <math.h>
+#include <stdlib.h>
+#include <string.h>
+
+// The default salt value.
+// This is the expansion of e (Euler's number) - the 19 digits after 2.71:
+// 8281828459045235360.
+// The above in base 256, from MSB to LSB:
+BitSequence SWIF_saltValueChar[SWIF_HAIFA_SALT_SIZE] = {114, 238, 247, 26, 192, 28, 170, 160};
+
+// All the IVs here below were produced from the decimal digits of e's expansion.
+// The code can be found in 'ProduceRandomIV.c'.
+// The initial value for 224 digest size.
+const BitSequence SWIF_HAIFA_IV_224[SWIFFTX_OUTPUT_BLOCK_SIZE] = 
+{37, 242, 132,   2, 167,  81, 158, 237, 113,  77, 162,  60,  65, 236, 108, 246,
+101,  72, 190, 109,  58, 205,  99,   6, 114, 169, 104, 114,  38, 146, 121, 142,
+ 59,  98, 233,  84,  72, 227,  22, 199,  17, 102, 198, 145,  24, 178,  37,   1,
+215, 245,  66, 120, 230, 193, 113, 253, 165, 218,  66, 134,  49, 231, 124, 204,
+  0};
+
+// The initial value for 256 digest size.
+const BitSequence SWIF_HAIFA_IV_256[SWIFFTX_OUTPUT_BLOCK_SIZE] = 
+{250,  50,  42,  40,  14, 233,  53,  48, 227,  42, 237, 187, 211, 120, 209, 234,
+  27, 144,   4,  61, 243, 244,  29, 247,  37, 162,  70,  11, 231, 196,  53,   6,
+ 193, 240,  94, 126, 204, 132, 104,  46, 114,  29,   3, 104, 118, 184, 201,   3,
+  57,  77,  91, 101,  31, 155,  84, 199, 228,  39, 198,  42, 248, 198, 201, 178,
+   8};
+
+// The initial value for 384 digest size.
+const BitSequence SWIF_HAIFA_IV_384[SWIFFTX_OUTPUT_BLOCK_SIZE] = 
+{40, 145, 193, 100, 205, 171,  47,  76, 254,  10, 196,  41, 165, 207, 200,  79,
+109,  13,  75, 201,  17, 172,  64, 162, 217,  22,  88,  39,  51,  30, 220, 151,
+133,  73, 216, 233, 184, 203,  77,   0, 248,  13,  28, 199,  30, 147, 232, 242,
+227, 124, 169, 174,  14,  45,  27,  87, 254,  73,  68, 136, 135, 159,  83, 152,
+  0};
+
+// The initial value for 512 digest size.
+const BitSequence SWIF_HAIFA_IV_512[SWIFFTX_OUTPUT_BLOCK_SIZE] = 
+{195, 126, 197, 167, 157, 114,  99, 126, 208, 105, 200,  90,  71, 195, 144, 138,
+ 142, 122, 123, 116,  24, 214, 168, 173, 203, 183, 194, 210, 102, 117, 138,  42,
+ 114, 118, 132,  33,  35, 149, 143, 163, 163, 183, 243, 175,  72,  22, 201, 255,
+ 102, 243,  22, 187, 211, 167, 239,  76, 164,  70,  80, 182, 181, 212,   9, 185,
+   0};
+
+
+///////////////////////////////////////////////////////////////////////////////////////////////
+// NIST API implementation portion.
+///////////////////////////////////////////////////////////////////////////////////////////////
+
+int Swifftx::Init(int hashbitlen)
+{
+	switch(hashbitlen)
+	{
+	case 224:
+		swifftxState.hashbitlen = hashbitlen;
+		// Initializes h_0 in HAIFA:
+		memcpy(swifftxState.currOutputBlock, SWIF_HAIFA_IV_224, SWIFFTX_OUTPUT_BLOCK_SIZE);
+		break;
+	case 256:
+		swifftxState.hashbitlen = hashbitlen;
+		memcpy(swifftxState.currOutputBlock, SWIF_HAIFA_IV_256, SWIFFTX_OUTPUT_BLOCK_SIZE);
+		break;
+	case 384:
+		swifftxState.hashbitlen = hashbitlen;
+		memcpy(swifftxState.currOutputBlock, SWIF_HAIFA_IV_384, SWIFFTX_OUTPUT_BLOCK_SIZE);
+		break;
+	case 512:
+		swifftxState.hashbitlen = hashbitlen;
+		memcpy(swifftxState.currOutputBlock, SWIF_HAIFA_IV_512, SWIFFTX_OUTPUT_BLOCK_SIZE);
+		break;
+	default:
+		return BAD_HASHBITLEN;
+	}
+	
+	swifftxState.wasUpdated = false;
+	swifftxState.remainingSize = 0;
+	memset(swifftxState.remaining, 0, SWIF_HAIFA_INPUT_BLOCK_SIZE);
+	memset(swifftxState.numOfBitsChar, 0, SWIF_HAIFA_NUM_OF_BITS_SIZE);
+	// Initialize the salt with the default value.
+	memcpy(swifftxState.salt, SWIF_saltValueChar, SWIF_HAIFA_SALT_SIZE);
+
+	InitializeSWIFFTX();
+
+	return SUCCESS;
+}
+
+int Swifftx::Update(const BitSequence *data, DataLength databitlen)
+{
+	// The size of input in bytes after putting the remaining data from previous invocation.
+	int sizeOfInputAfterRemaining = 0;
+	// The input block to compression function of SWIFFTX:
+	BitSequence currInputBlock[SWIFFTX_INPUT_BLOCK_SIZE] = {0};
+	// Whether we handled a single block.
+	bool wasSingleBlockHandled = false;
+
+	swifftxState.wasUpdated = true;
+
+	// Handle an empty message as required by NIST. Since 'Final()' is oblivious to the input
+	// (but of course uses the output of the compression function from the previous round, 
+	// which is called h_{i-1} in HAIFA article), we have to do nothing here.
+	if (databitlen == 0)
+		return SUCCESS;
+
+    // If we had before an input with unaligned length, return an error
+    if (swifftxState.remainingSize % 8)
+	{
+    	return INPUT_DATA_NOT_ALIGNED;
+    }
+
+    // Convert remaining size to bytes.
+    swifftxState.remainingSize /= 8;
+
+	// As long as we have enough data combined from (remaining + data) to fill input block
+	//NASTAVENIE RUND
+	while (((databitlen / 8) + swifftxState.remainingSize) >= SWIF_HAIFA_INPUT_BLOCK_SIZE)
+	{
+		// Fill the input block with data:
+		// 1. The output of the previous block:
+		memcpy(currInputBlock, swifftxState.currOutputBlock, SWIFFTX_OUTPUT_BLOCK_SIZE);
+		// 2. The input part of the block:
+		// 2a. The remaining data from the previous 'Update()' call:
+		if (swifftxState.remainingSize)
+			memcpy(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE, swifftxState.remaining, 
+				   swifftxState.remainingSize);
+		// 2b. The input data that we have place for after the 'remaining':
+		sizeOfInputAfterRemaining = SWIFFTX_INPUT_BLOCK_SIZE - SWIFFTX_OUTPUT_BLOCK_SIZE 
+								  - ((int) swifftxState.remainingSize) - SWIF_HAIFA_NUM_OF_BITS_SIZE 
+								  - SWIF_HAIFA_SALT_SIZE;
+		memcpy(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE + swifftxState.remainingSize, 
+			   data, sizeOfInputAfterRemaining);
+
+		// 3. The #bits part of the block:
+		memcpy(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE + swifftxState.remainingSize 
+			 + sizeOfInputAfterRemaining,
+			   swifftxState.numOfBitsChar, SWIF_HAIFA_NUM_OF_BITS_SIZE);
+		// 4. The salt part of the block:
+		memcpy(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE + swifftxState.remainingSize 
+			 + sizeOfInputAfterRemaining + SWIF_HAIFA_NUM_OF_BITS_SIZE,
+			   swifftxState.salt, SWIF_HAIFA_SALT_SIZE);
+
+		ComputeSingleSWIFFTX(currInputBlock, swifftxState.currOutputBlock, false);
+
+		// Update the #bits field with SWIF_HAIFA_INPUT_BLOCK_SIZE.
+		AddToCurrInBase256(swifftxState.numOfBitsChar, SWIF_HAIFA_INPUT_BLOCK_SIZE * 8);
+		wasSingleBlockHandled = true;
+		data += sizeOfInputAfterRemaining;
+		databitlen -= (sizeOfInputAfterRemaining * 8);
+   		swifftxState.remainingSize = 0;
+	}
+
+	// Update the swifftxState.remaining and swifftxState.remainingSize.
+    // remainingSize will be in bits after exiting 'Update()'.
+	if (wasSingleBlockHandled)
+	{		
+		swifftxState.remainingSize = (unsigned int) databitlen; // now remaining size is in bits.
+        if (swifftxState.remainingSize)
+			memcpy(swifftxState.remaining, data, (swifftxState.remainingSize + 7) / 8);
+	}
+	else
+	{
+		memcpy(swifftxState.remaining + swifftxState.remainingSize, data, 
+			   (size_t) (databitlen + 7) / 8);
+		swifftxState.remainingSize = (swifftxState.remainingSize * 8) + (unsigned short) databitlen;
+	}
+
+	return SUCCESS;
+}
+
+int Swifftx::Final(BitSequence *hashval)
+{
+    int i;
+    // Whether to add one last block. True if the padding appended to the last block overflows
+	// the block size.
+    bool toAddFinalBlock = false;
+    bool toPutOneInFinalBlock = false;
+    unsigned short oneShift = 0;
+   	// The size of the last input block before the zeroes padding. We add 1 here because we
+    // include the final '1' bit in the calculation and 7 as we round the length to bytes.
+	unsigned short sizeOfLastInputBlock = (swifftxState.remainingSize + 1 + 7) / 8;
+    // The number of bytes of zero in the padding part.
+	// The padding contains:
+	// 1. A single 1 bit.
+	// 2. As many zeroes as needed.
+	// 3. The message length in bits. Occupies SWIF_HAIFA_NUM_OF_BITS_SIZE bytes.
+	// 4. The digest size. Maximum is 512, so we need 2 bytes.
+	// If the total number achieved is negative, add an additional block, as HAIFA specifies.
+	short numOfZeroBytesInPadding = (short) SWIFFTX_INPUT_BLOCK_SIZE - SWIFFTX_OUTPUT_BLOCK_SIZE 
+								  - sizeOfLastInputBlock - (2 * SWIF_HAIFA_NUM_OF_BITS_SIZE) - 2 
+								  - SWIF_HAIFA_SALT_SIZE;
+   	// The input block to compression function of SWIFFTX:
+	BitSequence currInputBlock[SWIFFTX_INPUT_BLOCK_SIZE] = {0};
+	// The message length in base 256.
+	BitSequence messageLengthChar[SWIF_HAIFA_NUM_OF_BITS_SIZE] = {0};
+   	// The digest size used for padding:
+	unsigned char digestSizeLSB = swifftxState.hashbitlen % 256;
+	unsigned char digestSizeMSB = (swifftxState.hashbitlen - digestSizeLSB) / 256;
+
+	if (numOfZeroBytesInPadding < 1)
+		toAddFinalBlock = true;
+
+	// Fill the input block with data:
+	// 1. The output of the previous block:
+	memcpy(currInputBlock, swifftxState.currOutputBlock, SWIFFTX_OUTPUT_BLOCK_SIZE);
+	// 2a. The input part of the block, which is the remaining data from the previous 'Update()'
+    //     call, if exists and an extra '1' bit (maybe all we have is this extra 1):
+
+    // Add the last 1 in big-endian convention ...
+    if (swifftxState.remainingSize % 8 == 0)
+	{
+       swifftxState.remaining[sizeOfLastInputBlock - 1] = 0x80;
+    }
+    else 
+	{
+       swifftxState.remaining[sizeOfLastInputBlock - 1] |= (1 << (7 - (swifftxState.remainingSize % 8)));
+    }
+
+	if (sizeOfLastInputBlock)
+		memcpy(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE, swifftxState.remaining, 
+			   sizeOfLastInputBlock);
+    
+   	// Compute the message length in base 256:
+	for (i = 0; i < SWIF_HAIFA_NUM_OF_BITS_SIZE; ++i)
+        messageLengthChar[i] = swifftxState.numOfBitsChar[i];
+    if (sizeOfLastInputBlock)
+		AddToCurrInBase256(messageLengthChar, sizeOfLastInputBlock * 8);
+
+	if (!toAddFinalBlock)
+	{
+		// 2b. Put the zeroes:
+		memset(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE + sizeOfLastInputBlock,
+			   0, numOfZeroBytesInPadding);
+		// 2c. Pad the message length:
+		for (i = 0; i < SWIF_HAIFA_NUM_OF_BITS_SIZE; ++i)
+			currInputBlock[SWIFFTX_OUTPUT_BLOCK_SIZE + sizeOfLastInputBlock 
+						 + numOfZeroBytesInPadding + i] = messageLengthChar[i];
+		// 2d. Pad the digest size:
+		currInputBlock[SWIFFTX_OUTPUT_BLOCK_SIZE + sizeOfLastInputBlock 
+					 + numOfZeroBytesInPadding + SWIF_HAIFA_NUM_OF_BITS_SIZE] = digestSizeMSB;
+		currInputBlock[SWIFFTX_OUTPUT_BLOCK_SIZE + sizeOfLastInputBlock 
+					 + numOfZeroBytesInPadding + SWIF_HAIFA_NUM_OF_BITS_SIZE + 1] = digestSizeLSB;
+	}
+	else
+	{
+		// 2b. Put the zeroes, if at all:
+		if ((SWIF_HAIFA_INPUT_BLOCK_SIZE - sizeOfLastInputBlock) > 0)
+		{
+			 memset(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE + sizeOfLastInputBlock,
+					0, SWIF_HAIFA_INPUT_BLOCK_SIZE - sizeOfLastInputBlock);
+		}
+	}
+
+   	// 3. The #bits part of the block: 
+	memcpy(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE + SWIF_HAIFA_INPUT_BLOCK_SIZE, 
+           swifftxState.numOfBitsChar, SWIF_HAIFA_NUM_OF_BITS_SIZE);
+	// 4. The salt part of the block:
+	memcpy(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE + SWIF_HAIFA_INPUT_BLOCK_SIZE 
+		 + SWIF_HAIFA_NUM_OF_BITS_SIZE, 
+           swifftxState.salt, 
+		   SWIF_HAIFA_SALT_SIZE);
+
+    ComputeSingleSWIFFTX(currInputBlock, swifftxState.currOutputBlock, !toAddFinalBlock); 
+
+	// If we have to add one more block, it is now:
+	if (toAddFinalBlock)
+	{
+		// 1. The previous output block, as usual.
+		memcpy(currInputBlock, swifftxState.currOutputBlock, SWIFFTX_OUTPUT_BLOCK_SIZE);
+
+		// 2a. Instead of the input, zeroes:
+		memset(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE , 0, 
+			   SWIF_HAIFA_INPUT_BLOCK_SIZE - SWIF_HAIFA_NUM_OF_BITS_SIZE - 2);
+		// 2b. Instead of the input, the message length:
+		memcpy(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE + SWIF_HAIFA_INPUT_BLOCK_SIZE 
+			 - SWIF_HAIFA_NUM_OF_BITS_SIZE - 2,
+			   messageLengthChar,
+			   SWIF_HAIFA_NUM_OF_BITS_SIZE);
+		// 2c. Instead of the input, the digest size:
+		currInputBlock[SWIFFTX_OUTPUT_BLOCK_SIZE + SWIF_HAIFA_INPUT_BLOCK_SIZE - 2] = digestSizeMSB;
+		currInputBlock[SWIFFTX_OUTPUT_BLOCK_SIZE + SWIF_HAIFA_INPUT_BLOCK_SIZE - 1] = digestSizeLSB;
+		// 3. The #bits part of the block, which is zero in case of additional block:
+		memset(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE + SWIF_HAIFA_INPUT_BLOCK_SIZE,
+			   0,
+			   SWIF_HAIFA_NUM_OF_BITS_SIZE);
+		// 4. The salt part of the block:
+		memcpy(currInputBlock + SWIFFTX_OUTPUT_BLOCK_SIZE + SWIF_HAIFA_INPUT_BLOCK_SIZE 
+			 + SWIF_HAIFA_NUM_OF_BITS_SIZE, 
+               swifftxState.salt, 
+			   SWIF_HAIFA_SALT_SIZE);
+
+        ComputeSingleSWIFFTX(currInputBlock, swifftxState.currOutputBlock, true); 
+	}
+
+	// Finally, copy the result into 'hashval'. In case the digest size is not 512bit, copy the
+	// first hashbitlen of them:
+    for (i = 0; i < (swifftxState.hashbitlen / 8); ++i)
+		hashval[i] = swifftxState.currOutputBlock[i];
+
+	return SUCCESS;
+}
+
+int Swifftx::Hash(int hashbitlen, const BitSequence *data, DataLength databitlen, 
+				BitSequence *hashval)
+{
+	int result;
+	//hashState state;
+   	// The pointer to the current place in the input we take into the compression function.
+	DataLength currInputIndex = 0;
+
+    result = Swifftx::Init(hashbitlen);
+
+	if (result != SUCCESS)
+		return result;
+
+	for ( ; (databitlen / 8) >  SWIF_HAIFA_INPUT_BLOCK_SIZE; 
+         currInputIndex += SWIF_HAIFA_INPUT_BLOCK_SIZE, databitlen -= (SWIF_HAIFA_INPUT_BLOCK_SIZE * 8))
+	{
+		result = Swifftx::Update(data + currInputIndex, SWIF_HAIFA_INPUT_BLOCK_SIZE * 8); 
+		if (result != SUCCESS)
+			return result;
+	}
+
+	// The length of the last block may be shorter than (SWIF_HAIFA_INPUT_BLOCK_SIZE * 8)
+	result = Swifftx::Update(data + currInputIndex, databitlen); 
+	if (result != SUCCESS)
+	{
+		return result;
+	}
+
+    return Swifftx::Final(hashval);
+}
+
+///////////////////////////////////////////////////////////////////////////////////////////////
+// Helper fuction implementation portion.
+///////////////////////////////////////////////////////////////////////////////////////////////
+
+void Swifftx::AddToCurrInBase256(BitSequence value[SWIF_HAIFA_NUM_OF_BITS_SIZE], 
+							   unsigned short toAdd)
+{
+	unsigned char remainder = 0;
+	short i;
+	BitSequence currValueInBase256[8] = {0};
+	unsigned short currIndex = 7;
+	unsigned short temp = 0;
+
+	do
+	{
+		remainder = toAdd % 256;
+		currValueInBase256[currIndex--] = remainder;
+		toAdd -= remainder;
+		toAdd /= 256;
+	}
+	while(toAdd != 0);
+
+	for (i = 7; i >= 0; --i)
+	{
+		temp = value[i] + currValueInBase256[i];
+		if (temp > 255)
+		{
+			value[i] = temp % 256;
+			currValueInBase256[i - 1]++;
+		}
+		else
+			value[i] = (unsigned char) temp;
+	}
+}

algo/swifftx/Swifftx_sha3.h

@@ -0,0 +1,79 @@
+#ifndef SWIFFTX_SHA3_H
+#define SWIFFTX_SHA3_H
+
+#include "sha3_interface.h"
+#include "stdbool.h"
+#include "stdint.h"
+
+class Swifftx : public SHA3 {
+
+#define SWIFFTX_INPUT_BLOCK_SIZE 256
+#define SWIFFTX_OUTPUT_BLOCK_SIZE 65
+#define SWIF_HAIFA_SALT_SIZE 8
+#define SWIF_HAIFA_NUM_OF_BITS_SIZE 8
+#define SWIF_HAIFA_INPUT_BLOCK_SIZE (SWIFFTX_INPUT_BLOCK_SIZE - SWIFFTX_OUTPUT_BLOCK_SIZE \
+							  - SWIF_HAIFA_NUM_OF_BITS_SIZE - SWIF_HAIFA_SALT_SIZE)
+
+	typedef unsigned char BitSequence;
+//const DataLength SWIF_SALT_VALUE;
+
+#define SWIF_HAIFA_IV 0
+
+/*const BitSequence SWIF_HAIFA_IV_224[SWIFFTX_OUTPUT_BLOCK_SIZE];
+const BitSequence SWIF_HAIFA_IV_256[SWIFFTX_OUTPUT_BLOCK_SIZE];
+const BitSequence SWIF_HAIFA_IV_384[SWIFFTX_OUTPUT_BLOCK_SIZE];
+const BitSequence SWIF_HAIFA_IV_512[SWIFFTX_OUTPUT_BLOCK_SIZE];*/
+
+typedef enum 
+{ 
+	SUCCESS = 0,
+	FAIL = 1,
+	BAD_HASHBITLEN = 2,
+	BAD_SALT_SIZE = 3,
+	SET_SALT_VALUE_FAILED = 4,
+	INPUT_DATA_NOT_ALIGNED = 5
+} HashReturn;
+
+typedef struct hashState {
+	unsigned short hashbitlen;
+
+	// The data remained after the recent call to 'Update()'. 
+	BitSequence remaining[SWIF_HAIFA_INPUT_BLOCK_SIZE + 1];
+
+	// The size of the remaining data in bits.
+	// Is 0 in case there is no remaning data at all.
+	unsigned int remainingSize;
+
+	// The current output of the compression function. At the end will contain the final digest
+	// (which may be needed to be truncated, depending on hashbitlen).
+	BitSequence currOutputBlock[SWIFFTX_OUTPUT_BLOCK_SIZE];
+
+	// The value of '#bits hashed so far' field in HAIFA, in base 256.
+	BitSequence numOfBitsChar[SWIF_HAIFA_NUM_OF_BITS_SIZE];
+
+	// The salt value currently in use:
+	BitSequence salt[SWIF_HAIFA_SALT_SIZE];
+
+	// Indicates whether a single 'Update()' occured. 
+	// Ater a call to 'Update()' the key and the salt values cannot be changed.
+	bool wasUpdated;
+} hashState;
+
+private:
+int swifftxNumRounds;
+hashState swifftxState;
+
+
+public:
+int Init(int hashbitlen);
+int Update(const BitSequence *data, DataLength databitlen);
+int Final(BitSequence *hashval);
+int Hash(int hashbitlen, const BitSequence *data, DataLength databitlen, 
+				BitSequence *hashval);
+
+private:
+static void AddToCurrInBase256(BitSequence value[SWIF_HAIFA_NUM_OF_BITS_SIZE], unsigned short toAdd);
+
+};
+
+#endif

algo/swifftx/hash_interface.h

@@ -0,0 +1,21 @@
+#pragma once
+
+#include <cstdint>
+
+namespace hash {
+
+using BitSequence = unsigned char;
+using DataLength = unsigned long long;
+
+struct hash_interface {
+    virtual ~hash_interface() = default;
+
+    virtual int Init(int hash_bitsize) = 0;
+    virtual int Update(const BitSequence *data, DataLength data_bitsize) = 0;
+    virtual int Final(BitSequence *hash) = 0;
+
+    virtual int
+    Hash(int hash_bitsize, const BitSequence *data, DataLength data_bitsize, BitSequence *hash) = 0;
+};
+
+} // namespace hash

algo/swifftx/inttypes.h

@@ -0,0 +1,39 @@
+ /*
+  inttypes.h
+
+  Contributors:
+    Created by Marek Michalkiewicz <marekm@linux.org.pl>
+
+  THIS SOFTWARE IS NOT COPYRIGHTED
+
+  This source code is offered for use in the public domain.  You may
+ use, modify or distribute it freely.
+
+ This code is distributed in the hope that it will be useful, but
+ WITHOUT ANY WARRANTY.  ALL WARRANTIES, EXPRESS OR IMPLIED ARE HEREBY
+ DISCLAIMED.  This includes but is not limited to warranties of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+*/
+
+ #ifndef __INTTYPES_H_
+ #define __INTTYPES_H_
+
+ /* Use [u]intN_t if you need exactly N bits.
+  XXX - doesn't handle the -mint8 option.  */
+
+ typedef signed char swift_int8_t;
+ typedef unsigned char swift_uint8_t;
+
+ typedef int swift_int16_t;
+ typedef unsigned int swift_uint16_t;
+
+ typedef long swift_int32_t;
+ typedef unsigned long swift_uint32_t;
+
+ typedef long long swift_int64_t;
+ typedef unsigned long long swift_uint64_t;
+
+ //typedef swift_int16_t intptr_t;
+ //typedef swift_uint16_t uintptr_t;
+
+ #endif