v3.7.5

JayDDee · Dec 8, 2017 · af1c940 · af1c940
1 parent 4b57ac0
commit af1c940
Show file tree

Hide file tree

Showing 53 changed files with 1,297 additions and 4,763 deletions.
diff --git a/AUTHORS b/AUTHORS
@@ -16,4 +16,16 @@ LucasJones
 
 tpruvot@github
 
+elmad
+
+djm34
+
+palmd
+
+ig0tik3d
+
+Wolf0
+
+Optiminer
+
 Jay D Dee
diff --git a/Makefile.am b/Makefile.am
@@ -104,7 +104,9 @@ cpuminer_SOURCES = \
   algo/lyra2/sponge.c \
   algo/lyra2/lyra2rev2.c \
   algo/lyra2/lyra2re.c \
-  algo/lyra2/zcoin.c \
+  algo/lyra2/lyra2z-gate.c \
+  algo/lyra2/lyra2z.c \
+  algo/lyra2/lyra2z-4way.c \
   algo/lyra2/lyra2z330.c \
   algo/m7m.c \
   algo/neoscrypt.c \

diff --git a/README.md b/README.md
@@ -35,8 +35,9 @@ Supported Algorithms
                           heavy        Heavy
                           hmq1725      Espers
                           hodl         Hodlcoin
-                          jha          jackpotcoin
-                          keccak       Keccak
+                          jha          Jackpotcoin
+                          keccak       Maxcoin
+                          keccakc      Creative coin
                           lbry         LBC, LBRY Credits
                           luffa        Luffa
                           lyra2re      lyra2
@@ -50,7 +51,7 @@ Supported Algorithms
                           pentablake   Pentablake
                           phi1612      phi, LUX coin
                           pluck        Pluck:128 (Supcoin)
-                          polytimos
+                          polytimos    Ninja
                           quark        Quark
                           qubit        Qubit
                           scrypt       scrypt(1024, 1, 1) (default)

diff --git a/RELEASE_NOTES b/RELEASE_NOTES
@@ -164,6 +164,17 @@ Support for even older x86_64 without AES_NI or SSE2 is not availble.
 Change Log
 ----------
 
+v3.7.5
+
+New algo keccakc for Creative coin with 4way optimizations
+
+Rewrote some AVX/AVX2 code for more consistent implementation and some
+optimizing.
+
+Enhanced capabilities check to support 4way, mor eprecise reporting of
+features (not all algos use SSE2), and better error messages when using
+an incompatible pre-built version (Windows users).
+
 v3.7.4
 
 Removed unnecessary build options.

diff --git a/algo-gate-api.c b/algo-gate-api.c
@@ -77,6 +77,12 @@ void algo_not_tested()
   applog(LOG_WARNING,"and bad things may happen. Use at your own risk.");
 }
 
+void four_way_not_tested()
+{
+  applog( LOG_WARNING,"Algo %s has not been tested using 4way. It may not", algo_names[opt_algo] );
+  applog( LOG_WARNING,"work or may be slower. Please report your results.");
+}
+
 void algo_not_implemented()
 {
   applog(LOG_ERR,"Algo %s has not been Implemented.",algo_names[opt_algo]);
@@ -124,7 +130,7 @@ void init_algo_gate( algo_gate_t* gate )
    gate->do_this_thread          = (void*)&return_true;
    gate->longpoll_rpc_call       = (void*)&std_longpoll_rpc_call;
    gate->stratum_handle_response = (void*)&std_stratum_handle_response;
-   gate->optimizations           = SSE2_OPT;
+   gate->optimizations           = EMPTY_SET;
    gate->ntime_index             = STD_NTIME_INDEX;
    gate->nbits_index             = STD_NBITS_INDEX;
    gate->nonce_index             = STD_NONCE_INDEX;
@@ -171,11 +177,12 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
      case ALGO_HODL:         register_hodl_algo        ( gate ); break;
      case ALGO_JHA:          register_jha_algo         ( gate ); break;
      case ALGO_KECCAK:       register_keccak_algo      ( gate ); break;
+     case ALGO_KECCAKC:      register_keccakc_algo     ( gate ); break;
      case ALGO_LBRY:         register_lbry_algo        ( gate ); break;
      case ALGO_LUFFA:        register_luffa_algo       ( gate ); break;
      case ALGO_LYRA2RE:      register_lyra2re_algo     ( gate ); break;
      case ALGO_LYRA2REV2:    register_lyra2rev2_algo   ( gate ); break;
-     case ALGO_LYRA2Z:       register_zcoin_algo       ( gate ); break;
+     case ALGO_LYRA2Z:       register_lyra2z_algo      ( gate ); break;
      case ALGO_LYRA2Z330:    register_lyra2z330_algo   ( gate ); break;
      case ALGO_M7M:          register_m7m_algo         ( gate ); break;
      case ALGO_MYR_GR:       register_myriad_algo      ( gate ); break;

diff --git a/algo-gate-api.h b/algo-gate-api.h
@@ -85,12 +85,13 @@
 
 typedef  uint32_t set_t;
 
-#define EMPTY_SET 0
-#define SSE2_OPT  1
-#define AES_OPT   2
-#define AVX_OPT   4
-#define AVX2_OPT  8
-#define SHA_OPT  16
+#define EMPTY_SET       0
+#define SSE2_OPT        1
+#define AES_OPT         2  
+#define AVX_OPT         4
+#define AVX2_OPT        8
+#define SHA_OPT      0x10
+#define FOUR_WAY_OPT 0x20
 
 // return set containing all elements from sets a & b
 inline set_t set_union ( set_t a, set_t b ) { return a | b; }
@@ -156,7 +157,7 @@ bool return_false();
 void *return_null();
 void algo_not_tested();
 void algo_not_implemented();
-
+void four_way_not_tested();
 
 // Warning: algo_gate.nonce_index should only be used in targetted code
 // due to different behaviours by different targets. The JR2 index uses an

diff --git a/algo/blake/blake-4way.c b/algo/blake/blake-4way.c
@@ -9,18 +9,18 @@
 
 void blakehash_4way(void *state, const void *input)
 {
-     uint32_t hash0[16] __attribute__ ((aligned (64)));
-     uint32_t hash1[16] __attribute__ ((aligned (64)));
-     uint32_t hash2[16] __attribute__ ((aligned (64)));
-     uint32_t hash3[16] __attribute__ ((aligned (64)));
-     uint32_t vhash[16*4] __attribute__ ((aligned (64)));
+     uint32_t vhash[4*4] __attribute__ ((aligned (64)));
+     uint32_t hash0[4] __attribute__ ((aligned (32)));
+     uint32_t hash1[4] __attribute__ ((aligned (32)));
+     uint32_t hash2[4] __attribute__ ((aligned (32)));
+     uint32_t hash3[4] __attribute__ ((aligned (32)));
      blake256_4way_context ctx;
 
      blake256_4way_init( &ctx );
      blake256_4way( &ctx, input, 16 );
      blake256_4way_close( &ctx, vhash );
 
-     m128_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, 512 );
+     mm_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, 256 );
 
      memcpy( state,    hash0, 32 );
      memcpy( state+32, hash1, 32 );
@@ -32,7 +32,7 @@ int scanhash_blake_4way( int thr_id, struct work *work, uint32_t max_nonce,
                          uint64_t *hashes_done )
 {
    uint32_t vdata[20*4] __attribute__ ((aligned (64)));
-   uint32_t hash[4*8] __attribute__ ((aligned (64)));
+   uint32_t hash[4*4] __attribute__ ((aligned (32)));
    uint32_t *pdata = work->data;
    uint32_t *ptarget = work->target;
    const uint32_t first_nonce = pdata[19];
@@ -49,7 +49,7 @@ int scanhash_blake_4way( int thr_id, struct work *work, uint32_t max_nonce,
    // we need big endian data...
    swab32_array( endiandata, pdata, 20 );
 
-   m128_interleave_4x32( vdata, endiandata, endiandata, endiandata,
+   mm_interleave_4x32( vdata, endiandata, endiandata, endiandata,
                          endiandata, 640 );
 
    uint32_t *noncep = vdata + 76;   // 19*4

diff --git a/algo/blake/blake-gate.c b/algo/blake/blake-gate.c
@@ -13,11 +13,12 @@ bool register_blake_algo( algo_gate_t* gate )
 //  gate->scanhash  = (void*)&scanhash_blake_8way;
 //  gate->hash      = (void*)&blakehash_8way;
 #if defined(BLAKE_4WAY)
-  gate->optimizations = SSE2_OPT | AVX_OPT;
+  four_way_not_tested();
+  gate->optimizations = FOUR_WAY_OPT;
   gate->scanhash  = (void*)&scanhash_blake_4way;
   gate->hash      = (void*)&blakehash_4way;
+  four_way_not_tested();
 #else
-  gate->optimizations = SSE2_OPT;
   gate->scanhash  = (void*)&scanhash_blake;
   gate->hash      = (void*)&blakehash;
 #endif